System and method for generation and use of radiation outcome prediction score in patients undergoing radiotherapy

Abstract

A system and method for generation and use of radiation outcome prediction (response & side effects) score for patients undergoing radiotherapy for various medical conditions wherein the score is a personalized score, provided by analyzing multiple parameters including the tumor specific, patient specific, gene specific and treatment planning specific parameter(s), during and post therapy.

Claims

1. A system comprising: a processor storing instructions in non-transitory memory that, when executed, cause the processor to: receive raw data from multiple sources for a patient undergoing a treatment comprising radiation therapy, wherein the multiple sources comprise one or more of electronic medical records, imaging data, genomics data, metabolomics data, anatomical data, physiological data, and data from treatment planning systems; convert and combine the raw data into standardized usable record structure comprising input parameters for further processing, wherein the input parameters comprise one or more of tumor specific variables of a tumor, patient specific parameters, gene specific parameters, physics variables, metabolic variables, radiation variables, and treatment planning specific parameters of the patient during therapy and post therapy; store standardized data into records; perform analytics to gain sights using one or more machine learning models trained by utilizing plurality of patient records; provide knowledge represented through the machine learning models to aid a physician suggesting specification of treatment parameters for the patient, wherein the treatment parameters comprise one or more of total radiation dose, dose per fraction, overall treatment time, fractionation, beam type, beam energy, physical arrangements, and planned doses to critical normal tissue, wherein the machine learning models comprise deep learning and artificial neural networks; present an overall cumulative risk score predicting a treatment response and side effects of the radiation therapy, wherein the overall cumulative risk score comprises compromising one or more of a disease recurrence score, a time to recurrence score, a recovery probability score, and an adverse reaction score for the treatment based on the input parameters, wherein the overall cumulative risk score is a personalized score summarizing survival, adverse events, and the treatment response of the patient; provide a probabilistic predicting system adapted to be interactive to explore an impact of adjusting the treatment parameters on the overall cumulative risk score configured to adjust the treatment parameters; obtain new treatment parameters via the probabilistic predicting system by the physician based on a real-time condition of the patient to optimize the treatment response, wherein the treatment response comprises a radiation dose to the tumor versus dose to normal tissue for tumor regression with the side effects being minimal to normal tissues; update the overall cumulative risk score by incorporating the new treatment parameters and the treatment response through a feedback system; specify the new treatment radiotherapy dosage parameters in a treatment plan for the patient; wherein the new treatment parameters are determined based on characteristics of the patient as per the knowledge gained by the machine learning models from past treatments data of the patient and wherein the new treatment parameters are obtained by one or more of increasing the total radiation dose, decreasing the total radiation dose, adjusting the fractionation, adjusting the dose per fraction, altering the overall treatment time, altering the beam type, adjusting the beam energy adjusting the physical arrangements, adjusting dose to the normal tissues, and incorporating additional therapies to managing the side effects and improve treatment outcomes; and wherein the system generates real-time, dynamic, comprehensive therapy outcome prediction score for the radiation therapy via the overall cumulative risk score for the patient.

2. The system as claimed in claim 1, wherein the overall cumulative risk score is obtained using one or more of i) adjuvant radiotherapy progression free survival model configured for calculating a risk probability for survival, ii) adjuvant radiotherapy overall survival model configured for determining a success of adjuvant radiotherapy, iii) neoadjuvant radiotherapy RECIST criteria-based probability of tumor regression model configured for determining a first indicator for success of neoadjuvant radiotherapy, iv) neoadjuvant radiotherapy progression free survival model configured for determining a second indicator for success of neoadjuvant radiotherapy, v) neoadjuvant radiotherapy symptom improvement model configured for determining a third indicator for success of neoadjuvant radiotherapy, vi) radical radiotherapy overall survival model configured for determining a first indicator for success of radical radiotherapy, vii) radical radiotherapy RECIST criteria-based probability of tumor regression model configured for determining a second indicator for success of radical radiotherapy, viii) radical radiotherapy progression free survival model configured for determining a third indicator for success of radical radiotherapy, ix) radical radiotherapy symptom improvement model configured for determining a fourth indicator for success of radical radiotherapy, x) palliative radiotherapy symptom improvement model configured for determining a first indicator for success of palliative radiotherapy, xi) palliative radiotherapy RECIST criteria-based probability of tumor regression model configured for determining a second indicator for success of palliative radiotherapy, xii) palliative radiotherapy progression free survival model configured for determining a third indicator for success of palliative radiotherapy, xiii) palliative radiotherapy overall survival model configured for determining a fourth indicator for success of palliative radiotherapy, xiv) radiotherapy areas of recurrence model configured for determining an area of recurrence following radiotherapy, xv) radiotherapy side effects grade model configured to predict a grade of the side effects following radiotherapy, xvi) radiotherapy side effects time to resolution model configured to determine a time to resolution of the side effects post radiotherapy, and xvii) radiotherapy non-resolving side effects model configured to predict the side effects that will not resolve post radiotherapy.

3. The system as claimed in claim 1, wherein the system is configured to learn from the feedback system, wherein a feedback comprises tumor response and toxicity; and wherein the feedback is utilized to re-calibrate the overall cumulative risk score and suggest timely medical intervention based on the overall cumulative risk score.

4. The system as claimed in claim 1, wherein the feedback system is used to further re-calibrates the overall cumulative risk score for the treatment response and toxicity, based on one or more of tumor regression, oxygen concentration, and the side effects and update a radiation plan.

5. The system as claimed in claim 1, wherein the machine learning models capture an interplay of the input parameters that potentially influence the treatment response by virtue of cellular, tissue, and organ level makeup of the tumor and interaction of the tumor with the normal tissues.

6. The system as claimed in claim 1, wherein the input parameters further comprises demographic variables and social variables.

7. The system as claimed in claim 1, wherein the system is configured to provide predict of a response of the tumor to the radiation therapy and toxicity to the normal tissues.

8. The system as claimed in claim 1, wherein a change in one of the input parameters or the change in one of the input parameters in combination with other input parameters is used to understand an interplay of the input parameters.

9. The system of claim 1, wherein the patient specific parameters comprise one or more of age, smoking history, alcohol use, ethnicity, weight, height, volume of organ proportionate to tumor, genetic makeup, body fat, food habits, co-morbidity data comprising one or more of diabetes, collagen vascular disease, hypertension, inflammatory diseases.

10. The system of claim 1, wherein the tumor specific variables comprise one or more of type of tumor, stage of tumor, node, metastases (TNM), pathology, volume, vascular density, oxygenation, hydration status, genetic makeup comprising one or more of ER, PR, P53, HER 2, grade of the tumor, and Ki-67 index.

11. The system of claim 1, wherein the gene specific parameters comprise gene variations of one or more of irs1 (XRCC2), irs2 (XRCC8), irs3 (RAD51C), irs20 (PRKDC), IRS1-SF (XRCC3), xrs5 (XRCC5) and XR-1 (XRCC4).

12. The system of claim 1, wherein the physics variables comprise one or more of total radiation dose, dose per fraction, fractionation, overall treatment time, planned doses to the normal tissues, temperature of organ, temperature of a therapy room, and beam features.

13. The system of claim 1, wherein the metabolic variables comprise one or more of lipid content, bone density, muscle mass, hypoxic tissue, estimated pH, hydration status, possibility of pre-existing free-radical insult at cellular level.

14. The system of claim 6, wherein the radiation variables comprise one or more of a beam strength and a beam contour.

15. The system of claim 1, wherein the machine learning model further comprise one or more of Recurrent Neural Networks (RNN), Long Short Term Memory (LSTM), Convolution Neural Networks (CNN).

16. The system of claim 1, wherein the machine learning models comprise one or more of logistic regression model, linear regression model, support vector machines model, Classification and Regression Trees (CART) model, boosting model, bagging model, and random forests model.

17. The system of claim 1, wherein the system is further configured to determine an output comprising a probability of overall survival at a future time point due to the treatment.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 shows the detailed schematic representation of the device.

(2) FIG. 2 illustrates the system workflow of data processing and analysis

(3) FIG. 3 illustrates the architecture of the solution

(4) FIG. 4 illustrates the overall process of data analysis comprising the input layer, analytical and output layer.

DETAILED DESCRIPTION OF THE INVENTION

(5) The illustrative embodiments of the system include Artificial Intelligence based software system(s) for determining and analyzing radiotherapy treatment using tumor parameters. According to one embodiment, the tumor parameters shall be considered, that are known or postulated to affect the response rates of the cancer treatment using radio therapy. The system provides a method for computing scores that can be used to assess a cancer treatment specification for a given set of patient and tumor characteristics. Subsequently, the system aids physicians in determining optimal dose of radiation required to combat the tumors.

(6) In one embodiment, the system receives data about patient characteristics pertaining to clinical, medical, genomics, metabolomics and tumor information collected from his clinical reports, laboratory tests, Computed tomography (CT) scan and Magnetic Resonance imaging (MRI).

(7) Data about the tumors is not limited to size, organ of origin, tissue of origin (histopathology), immuno-histo-chemical behavior, grade of tumor, other known variables like lympho-vascular involvement, margins and other known factors as per availability. The system considers all the above data or only such data as may be relevant, on a case-to-case basis.

(8) According to one embodiment, the system considers patient clinical data related parameters that includes but are not limited to age, smoking history, alcohol use, ethnicity, weight, height, volume of organ proportionate to tumor, co-morbidity (for example, diabetes, collagen vascular disease, hypertension, inflammatory bowel disease), genetic makeup, body fat, food habits etc. shall be evaluated

(9) According to one embodiment, the system considers patient genomic data related parameters that includes but not limited an array of known and proposed gene variations-irs1 (XRCC2), irs2 (XRCC8), irs3 (RAD51C), irs20 (PRKDC), IRS1-SF (XRCC3), xrs5 (XRCC5) and XR-1 (XRCC4) and all those that offer sensitivity or resistance will be taken into account.

(10) According to one embodiment, the system considers patient metabolomic data related parameters that includes but are not limited to metabolic variables like lipid content/bone density/muscle mass, hypoxic tissue, estimated pH, hydration status, possibility of pre-existing free-radical insult at cellular level shall be considered

(11) According to one embodiment, the system considers the anatomical variables like distance of organ traversed in RT field, vessel wall thickness, hydration status of tissue, percent of necrosis, stromal effect, tissues surrounding the tumor, proximity to critical structures.

(12) According to one embodiment, the system considers physics variables like temp of organ vs room, beam features, dose fractions.

(13) According to one embodiment, the system considers the physiological variables like blood flow/local pH/oxygenation.

(14) According to one embodiment, the system takes into account the principles of radiation physics like total radiation dose, dose per fraction, overall treatment time, planned doses to critical normal tissues.

(15) According to one embodiment, the system considers, periodically, the physician and patient reported toxicity parameters.

(16) The entire data, changing from baseline and over a period of time shall be fed into the AI/Machine learning algorithm

(17) The system generates the scores for assessment of a treatment using radiation therapy as illustrated in FIG. 1 and FIG. 2 by leveraging a set of software modules. These modules are interface module, harmonization module, storage module, machine learning module, clinical decision module and treatment assessment module.

(18) The interface module 206 (also referred to as input layer 114 of FIG. 1) receives raw data about the patient's characteristics and a suggested radio therapy treatment to that patient from the physicians (shown as 130 of FIG. 1). The data about the patient medical characteristics 202 and treatment characteristics 204 can be collected from devices or manually entered by physicians or caretakers. This raw data from interface module is transmitted to harmonization module 208 (116 of FIG. 1), to combine the raw data captured from multiple sources into uniform, unambiguous, standardized usable record structure for further processing. The data from harmonization module is diffused to storage module 210 (118 of FIG. 1) for storage of standardized patient's data and treatment information into records. These records are to be utilized for training the machine learning models (shown as 124 in FIG. 1) and for performing analytics to gaining insight into the most appropriate treatment for a patient based on their clinical and medical characteristics. The Machine Learning Module 124 consumes data that has been preprocessed (using data preprocessing module shown as 122 in FIG. 1) and utilizes techniques such as deep learning and artificial neural networks including but not limited to Recurrent Neural Networks (RNN), Long Short Term Memory (LSTM), Convolution Neural Networks (CNN), traditional machine learning techniques including but not limited to linear models (logistic regression, linear regression, support vector machines), Classification and Regression Trees (CART), boosting, bagging, random forests and other statistical techniques and rule based algorithms. The Machine Learning Module 124 produces output includingthe probability of Progression Free Survival at time point t after Adjuvant Radiotherapy, the probability of Overall Survival at time point t after Adjuvant Radiotherapy, the probability of tumor regression after Neoadjuvant Radiotherapy, probability of Progression Free Survival at time t after Neoadjuvant radiotherapy, probability of symptom improvement after Neoadjuvant radiotherapy, probability of Overall Survival at time point t after Radical radiotherapy, probability of tumor regression after Radical radiotherapy, probability of Progression Free Survival at time t after Radical Radiotherapy, probability of symptom improvement after Radical Radiotherapy, probability of symptom improvement after Palliative Radiotherapy, probability of tumor regression after Palliative Radiotherapy, probability of Progression Free Survival at time t after Palliative Radiotherapy, probability of Overall Survival at time t after Palliative Radiotherapy, probability of tumor recurrence in organs or tissues after radiotherapy, probability of a particular grade of side effect occurring after radiotherapy, probability of resolution of side effect at time t after radiotherapy and probability of resolution of side effects after radiotherapy. Clinical decision module 212 (120 of FIG. 1) provide knowledge represented through machine learning models to aid the physicians suggesting optimal specification of treatment parameters for a patient. These models are trained on the data stored in storage module until they attain power to predict scores related to a specific radiotherapy treatment. When the trained models are presented with a patient's characteristics and treatment specification, they will generate raw scores.

(19) Output layer 214 (126 of FIG. 1), presents the disease recurrence score, time to recurrence score and adverse reaction score for a specific treatment. Theses scores provides a platform for assessment of the treatments by the physicians. Treatment assessment 216 module is used to ratify the radiotherapy dosage parameters. Whether a specific dosage of the give patient characteristics will be appropriate as per the knowledge the model gained form past patient-treatments data. The final output of the whole process is the precise specifications of the parameters of the treatment 218 are presented to physicians for further clinical actions.

(20) The Clinical decision module 212 has a set of models built to acquire insights about the assessment of radio therapy treatment from multiple aspects using scores. In one embodiment, the Clinical decision module has these models: 1. Adjuvant Radiotherapy Progression Free Survival (Time to recurrence) Model 2. Adjuvant Radiotherapy Overall Survival (Time to recurrence) Model 3. Neoadjuvant Radiotherapy RECIST criteria-based Probability of tumor regression Model 4. Neoadjuvant Radiotherapy Progression Free Survival (Time to recurrence) Model: 5. Neoadjuvant Radiotherapy Symptom Improvement Model: 6. Radical Radiotherapy Overall Survival (Time to recurrence) Model 7. Radical Radiotherapy RECIST criteria-based Probability of tumor regression Model: 8. Radical Radiotherapy Progression Free Survival (Time to recurrence) Model: 9. Radical Radiotherapy Symptom Improvement Model 10. Palliative Radiotherapy Symptom Improvement Model 11. Palliative Radiotherapy RECIST criteria-based Probability of tumor regression Model 12. Palliative Radiotherapy Progression Free Survival (Time to recurrence) Model 13. Palliative Radiotherapy Overall Survival (Time to recurrence) Model 14. Radiotherapy Areas of Recurrence Model 15. Radiotherapy Side effects Grade Model 16. Radiotherapy Side effects Time to Resolution 17. Radiotherapy Non-resolving Side Effects Model

(21) Each model has an objective/event for the give set of input characteristics and the generated outcome provides an information in terms of probability about the occurrence of the event with respect to time. Each model's objective and its characteristics are discussed below:

(22) 1. Adjuvant Radiotherapy Progression Free Survival (Time to Recurrence) Model:

(23) Progression Free Survival serves as the primary indicator for success of adjuvant radiotherapy.

(24) A model was built to determine Progression free survival after adjuvant radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 1.

(25) TABLE-US-00002 TABLE 1 List of variables considered for modelling the Adjuvant Radiotherapy Progression Free Survival Age Smoking history Alcohol use Ethnicity Weight Height Volume of organ/tissue Diabetes e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory bowel Body fat disease disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor pathology Tumor volume Tumor vascular density Tumor oxygenation (estimated by imaging) Tumor hydration Tumor marker (ER) Tumor marker (PR) Tumor marker (HER 2) status Tumor marker Ki-67 index Genetic variations irs1 Genetic variation irs2 (other) (XRCC2) (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall thickness hydration status of tissue traversed in RT field Percent of necrosis stromal effect tissues surrounding the proximity to critical tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ vs Room Beam features existing free-radical insult at cellular level Total radiation dose dose per fraction overall treatment time planned doses to critical normal tissues site-specific Physician-reported use of chemotherapy use of hormone therapy patient-reported toxicity toxicity use of surgery concurrent medications b. Independent Variable: Time to progression of disease c. Type of Model: Survival (Time to event) Model d. Output of Model: Probability of Progression free survival at time point t (t is a time point like 1 month after adjuvant radiotherapy).
2. Adjuvant Radiotherapy Overall Survival (Time to recurrence) Model:
Overall Survival serves as the secondary indicator for success of adjuvant radiotherapy.

(26) A model was built to determine Overall survival after adjuvant radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 2.

(27) TABLE-US-00003 TABLE 2 List of variables considered for modelling the Adjuvant Radiotherapy Overall Survival Age Smoking history Alcohol use Ethnicity Weight Height Volume of organ/tissue Diabetes e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory bowel Body fat disease disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor pathology Tumor volume Tumor vascular density Tumor oxygenation (estimated by imaging) Tumor hydration Tumor marker (ER) Tumor marker (PR) Tumor marker (HER 2) status Tumor marker Ki-67 index Genetic variations irs1 Genetic variation irs2 (other) (XRCC2) (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall thickness hydration status of tissue traversed in RT field Percent of necrosis stromal effect tissues surrounding the proximity to critical tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ vs Room Beam features existing free-radical insult at cellular level Total radiation dose dose per fraction overall treatment time planned doses to critical normal tissues site-specific Physician-reported use of chemotherapy use of hormone therapy patient-reported toxicity toxicity use of surgery concurrent medications b. Independent Variable: Time to overall survival c. Type of Model: Survival (Time to event) Model d. Output of Model: Probability of overall survival at time point t (t is a time point like 1 month after adjuvant radiotherapy)
3. Neoadjuvant Radiotherapy RECIST Criteria-Based Probability of Tumor Regression Model:

(28) RECIST criteria-based Probability of tumor regression serves as the primary indicator for success of neoadjuvant radiotherapy.

(29) A model was built to determine Probability of tumor regression after neoadjuvant radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 3.

(30) TABLE-US-00004 TABLE 3 List of variables considered for modelling Neoadjuvant Radiotherapy RECIST criteria-based Probability Age Smoking history Alcohol use Ethnicity Weight Height Volume of organ/tissue Diabetes e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory bowel Body fat disease disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor pathology Tumor volume Tumor vascular density Tumor oxygenation (estimated by imaging) Tumor hydration Tumor marker (ER) Tumor marker (PR) Tumor marker (HER 2) status Tumor marker Ki-67 index Genetic variations irs1 Genetic variation irs2 (other) (XRCC2) (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall thickness hydration status of tissue traversed in RT field Percent of necrosis stromal effect tissues surrounding the proximity to critical tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ vs Room Beam features existing free-radical insult at cellular level Total radiation dose dose per fraction overall treatment time planned doses to critical normal tissues site-specific Physician-reported use of chemotherapy use of hormone therapy patient-reported toxicity toxicity use of surgery concurrent medications b. Independent Variable: Whether tumor regression occurred? The mapping from RECIST criteria to whether tumor regression occurred was as follows: Data points with Complete Response and Partial Response were together considered as tumor regression occurred Data Points with Stable Disease and Progressive Disease were together considered as absence of tumor regression. c. Type of Model: Binary Classification Model d. Output of Model: Probability of tumor regression post neoadjuvant radiotherapy
Neoadjuvant Radiotherapy Progression Free Survival (Time to Recurrence) Model:

(31) Progression Free Survival serves as the secondary indicator for success of neoadjuvant radiotherapy.

(32) A model was built to determine Progression free survival after neoadjuvant radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 4.

(33) TABLE-US-00005 TABLE 4 List of variables considered for modelling Neoadjuvant Radiotherapy Progression Free Survival Age Smoking history Alcohol use Ethnicity Weight Height Volume of organ/tissue Diabetes e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory bowel Body fat disease disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor pathology Tumor volume Tumor vascular density Tumor oxygenation (estimated by imaging) Tumor hydration Tumor marker (ER) Tumor marker (PR) Tumor marker (HER 2) status Tumor marker Ki-67 index Genetic variations irsl Genetic variation irs2 (other) (XRCC2) (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall thickness hydration status of tissue traversed in RT field Percent of necrosis stromal effect tissues surrounding the proximity to critical tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ vs Room Beam features existing free-radical insult at cellular level Total radiation dose dose per fraction overall treatment time planned doses to critical normal tissues site-specific Physician-reported use of chemotherapy use of hormone therapy patient-reported toxicity toxicity use of surgery concurrent medications b. Independent Variable: Time to progression of disease c. Type of Model: Survival (Time to event) Model d. Output of Model: Probability of Progression free survival at time point t (t is a time point like 1 month after neoadjuvant radiotherapy)
5. Neoadjuvant Radiotherapy Symptom Improvement Model:

(34) Probability of Symptom Improvement serves as the tertiary indicator for success of neoadjuvant radiotherapy.

(35) A model was built to determine Probability of Symptom Improvement after neoadjuvant radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 5.

(36) TABLE-US-00006 TABLE 5 List of variables considered for modelling Neoadjuvant Radiotherapy Symptom Improvement Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory bowel Body fat disease disease Nutrition Tumor type Tumor stage Tumor grade (TNM) Tumor pathology Tumor volume Tumor vascular Tumor density oxygenation (estimated by imaging) Tumor hydration Tumor marker Tumor marker Tumor marker status (ER) (PR) (HER 2) Tumor marker Ki-67 index Genetic Genetic (other) variations variation irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF xrs5 (XRCC5) (XRCC3) XR-1 (XRCC4) Distance of organ Vessel wall hydration traversed in RT thickness status of field tissue Percent of stromal effect tissues proximity necrosis surrounding to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ Beam features existing free- vs Room radical insult at cellular level Total radiation dose per fraction overall planned doses to dose treatment critical normal time tissues site-specific Physician-reported use of use of hormone patient- toxicity chemotherapy therapy reported toxicity use of surgery concurrent medications b. Dependent Variable: Whether Symptom Improvement occurred? c. Type of Model: Binary Classification Model d. Output of Model: Probability of symptom improvement post neoadjuvant radiotherapy
6. Radical Radiotherapy Overall Survival (Time to Recurrence) Model:

(37) Overall Survival serves as the primary indicator for success of Radical radiotherapy. A model was built to determine Overall survival after Radical radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 6.

(38) TABLE-US-00007 TABLE 6 List of variables for modelling Radical Radiotherapy Overall Survival Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory Body fat disease bowel disease Nutrition Tumor type Tumor stage Tumor grade (TNM) Tumor Tumor volume Tumor vascular Tumor pathology density oxygenation (estimated by imaging) Tumor Tumor marker Tumor marker Tumor marker hydration (ER) (PR) (HER 2) status Tumor marker Ki-67 index Genetic variations Genetic variation (other) irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 Distance of organ Vessel wall hydration status (XRCC4) traversed in thickness of tissue RT field Percent of stromal effect tissues proximity necrosis surrounding to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ vs Beam features existing Room free-radical insult at cellular level Total radiation dose per fraction overall treatment planned doses to dose time critical normal tissues site-specific Physician- use of use of hormone patient-reported reported chemotherapy therapy toxicity toxicity use of surgery concurrent medications b. Independent Variable: Time to overall survival c. Type of Model: Survival (Time to event) Model

(39) d. Output of Model: Probability of overall survival at time point t (t is a time point like 1 month after Radical radiotherapy)

(40) 7. Radical Radiotherapy RECIST Criteria-Based Probability of Tumor Regression Model:

(41) RECIST criteria-based Probability of tumor regression serves as the secondary indicator for success of Radical radiotherapy.

(42) A model was built to determine Probability of tumor regression after Radical radiotherapy. a. Dependent Variables: Including but not limited to

(43) TABLE-US-00008 TABLE 7 List of variables considered for modelling Radical Radiotherapy RECIST criteria-based Probability Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory Body fat disease bowel disease Nutrition Tumor type Tumor stage Tumor grade (TNM) Tumor pathology Tumor volume Tumor vascular Tumor density oxygenation (estimated by imaging) Tumor hydration Tumor marker Tumor marker Tumor marker status (ER) (PR) (HER 2) Tumor marker Ki-67 index Genetic Genetic (other) variations variation irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF xrs5 (XRCC5) (XRCC3) XR-1 (XRCC4) Distance of organ Vessel wall hydration traversed in RT thickness status of field tissue Percent of stromal effect tissues proximity necrosis surrounding to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ Beam features existing free- vs Room radical insult at cellular level Total radiation dose per fraction overall planned doses to dose treatment critical normal time tissues site-specific Physician- use of use of hormone patient- reported chemotherapy therapy reported toxicity toxicity use of surgery concurrent medications b. Independent Variable: Whether tumor regression occurred?

(44) The mapping from RECIST criteria to whether tumor regression occurred was as follows:

(45) Data points with Complete Response and Partial Response were together considered as tumor regression occurred

(46) Data Points with Stable Disease and Progressive Disease were together considered as absence of tumor regression. c. Type of Model: Binary Classification Model d. Output of Model: Probability of tumor regression post Radical radiotherapy
8. Radical Radiotherapy Progression Free Survival (Time to Recurrence) Model:

(47) Progression Free Survival serves as the tertiary indicator for success of Radical radiotherapy.

(48) A model was built to determine Progression free survival after Radical radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics that

(49) TABLE-US-00009 TABLE 8 List of variables considered for modelling Radical Radiotherapy Progression Free Survival Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen Inflammatory Body fat vascular bowel disease disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor pathology Tumor volume Tumor vascular Tumor density oxygenation (estimated by imaging) Tumor hydration Tumor marker Tumor marker Tumor marker status (ER) (PR) (HER 2) Tumor marker Ki-67 index Genetic Genetic (other) variations variation irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall hydration traversed in RT thickness status of field tissue Percent of stromal effect tissues proximity necrosis surrounding to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ Beam features existing free-radical vs Room insult at cellular level Total radiation dose per fraction overall planned dose treatment doses to time critical normal tissues site-specific Physician- use of use of patient- reported chemotherapy hormone reported toxicity toxicity therapy use of surgery concurrent medications b. Independent Variable: Time to progression of disease c. Type of Model: Survival (Time to event) Model d. Output of Model: Probability of Progression free survival at time point t (t is a time point like 1 month after Radical radiotherapy)
9. Radical Radiotherapy Symptom Improvement Model:

(50) Probability of Symptom Improvement serves as the quaternary indicator for success of Radical radiotherapy.

(51) A model was built to determine Probability of Symptom Improvement after Radical radiotherapy. a. Dependent Variables: Including but not limited to

(52) TABLE-US-00010 Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory bowel Body fat disease disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor Tumor volume Tumor Tumor pathology vascular oxygenation density (estimated by imaging) Tumor hydration Tumor marker Tumor marker Tumor marker status (ER) (PR) (HER 2) Tumor marker Ki-67 index Genetic variations Genetic variation (other) irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance Vessel wall hydration of organ thickness status of traversed in RT tissue field Percent of stromal effect tissues proximity necrosis surrounding to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ Beam features existing free- vs Room radical insult at cellular level Total radiation dose per overall planned doses to dose fraction treatment critical normal time tissues site-specific Physician- use of use of hormone patient- reported chemotherapy therapy reported toxicity toxicity use of surgery concurrent medications b. Independent Variable: Whether Symptom Improvement occurred? c. Type of Model: Binary Classification Model d. Output of Model: Probability of symptom improvement post Radical radiotherapy
10. Palliative Radiotherapy Symptom Improvement Model:

(53) Probability of Symptom Improvement serves as the primary indicator for success of palliative radiotherapy.

(54) A model was built to determine Probability of Symptom Improvement after Palliative radiotherapy. a. Dependent Variables: Including but not limited to list of characteristics presented in Table 10.

(55) TABLE-US-00011 TABLE 10 List of variables considered for modelling Palliative Radiotherapy Symptom Improvement Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory bowel Body fat disease disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor pathology Tumor volume Tumor Tumor vascular oxygenation density (estimated by imaging) Tumor hydration Tumor marker Tumor marker Tumor marker status (ER) (PR) (HER 2) Tumor marker Ki-67 index Genetic Genetic (other) variations variation irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance Vessel wall hydration of organ thickness status of traversed in RT tissue field Percent of necrosis stromal effect tissues surrounding proximity to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ Beam features existing free-radical vs Room insult at cellular level Total radiation dose dose per fraction overall treatment planned doses to time critical normal tissues site-specific Physician- use of use of hormone patient- reported chemotherapy therapy reported toxicity toxicity use of surgery concurrent medications b. Independent Variable: Whether Symptom Improvement occurred? c. Type of Model: Binary Classification Model d. Output of Model: Probability of symptom improvement post Palliative radiotherapy
11. Palliative Radiotherapy RECIST Criteria-Based Probability of Tumor Regression Model:

(56) RECIST criteria-based Probability of tumor regression serves as the secondary indicator for success of Palliative radiotherapy.

(57) A model was built to determine Probability of tumor regression after Palliative radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics as presented in Table 11.

(58) TABLE-US-00012 TABLE 11 List of variables considered for modelling for Palliative Radiotherapy RECIST criteria-based Probability of tumor regression Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory Body fat disease bowel disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor Tumor volume Tumor vascular Tumor pathology density oxygenation (estimated by imaging) Tumor Tumor marker Tumor marker Tumor marker hydration (ER) (PR) (HER 2) status Tumor marker Ki-67 index Genetic Genetic variations variation (other) irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall hydration traversed thickness status in RT field of tissue Percent of stromal effect tissues proximity to necrosis surrounding critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ Beam features existing vs Room free-radical insult at cellular level Total radiation dose per fraction overall treatment planned doses to dose time critical normal tissues site-specific Physician-reported use of use of hormone patient- toxicity chemotherapy therapy reported toxicity use of surgery concurrent medications b. Independent Variable: Whether tumor regression occurred?

(59) The mapping from RECIST criteria to whether tumor regression occurred was as follows:

(60) Data points with Complete Response and Partial Response were together considered as tumor regression occurred

(61) Data Points with Stable Disease and Progressive Disease were together considered as absence of tumor regression. c. Type of Model: Binary Classification Model d. Output of Model: Probability of tumor regression post Palliative radiotherapy
12. Palliative Radiotherapy Progression Free Survival (Time to Recurrence) Model:

(62) Progression Free Survival serves as the tertiary indicator for success of Palliative radiotherapy.

(63) A model was built to determine Progression free survival after Palliative radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 12.

(64) TABLE-US-00013 TABLE 12 List of variables considered for modelling Palliative Radiotherapy Progression Free Survival Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory Body fat disease bowel disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor pathology Tumor volume Tumor vascular Tumor density oxygenation (estimated by imaging) Tumor hydration Tumor marker Tumor marker Tumor marker status (ER) (PR) (HER 2) Tumor marker Ki-67 index Genetic variations Genetic variation (other) irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall hydration status traversed in RT thickness of tissue field Percent of stromal effect tissues proximity necrosis surrounding to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ Beam features existing vs Room free-radical insult at cellular level Total radiation dose per fraction overall treatment planned doses to dose time critical normal tissues site-specific Physician- use of use of hormone patient- reported chemotherapy therapy reported toxicity toxicity use of surgery concurrent medications b. Independent Variable: Time to progression of disease c. Type of Model: Survival (Time to event) Model d. Output of Model: Probability of Progression free survival at time point t (t is a time point like 1 month after Palliative radiotherapy)
13. Palliative Radiotherapy Overall Survival (Time to Recurrence) Model:

(65) Overall Survival serves as the quaternary indicator for success of Palliative radiotherapy.

(66) A model was built to determine Overall survival after Palliative radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 13.

(67) TABLE-US-00014 TABLE 13 List of variables considered for modelling Palliative Radiotherapy Overall Survival Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory Body fat disease bowel disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor Tumor volume Tumor vascular Tumor pathology density oxygenation (estimated by imaging) Tumor hydration Tumor marker Tumor marker Tumor marker status (ER) (PR) (HER 2) Tumor marker Ki-67 index Genetic variations Genetic variation (other) irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall hydration status traversed in RT thickness of tissue field Percent of stromal effect tissues proximity necrosis surrounding to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ vs Beam features existing Room free-radical insult at cellular level Total radiation dose per fraction overall treatment planned doses to dose time critical normal tissues site-specific Physician- use of use of hormone patient-reported reported chemotherapy therapy toxicity toxicity use of surgery concurrent medications b. Independent Variable: Time to overall survival c. Type of Model: Survival (Time to event) Model d. Output of Model: Probability of overall survival at time point t (t is a time point like 1 month after Palliative radiotherapy)
14. Radiotherapy Areas of Recurrence Model: A model was built to predict areas of recurrence following radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 14.

(68) TABLE-US-00015 TABLE 14 List of the variables considered for modelling as Radiotherapy Areas of Recurrence Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory Body fat disease bowel disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor pathology Tumor volume Tumor vascular Tumor density oxygenation (estimated by imaging) Tumor hydration Tumor marker Tumor marker Tumor marker status (ER) (PR) (HER 2) Tumor marker Ki-67 index Genetic variations Genetic variation (other) irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall hydration traversed in RT thickness status of field tissue Percent of necrosis stromal effect tissues surrounding proximity to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of Temp of organ Beam features pre-existing vs Room free-radical insult at cellular level Total radiation dose per fraction overall treatment planned doses to dose time critical normal tissues site-specific Physician- use of use of hormone patient- reported chemotherapy therapy reported toxicity toxicity use of surgery concurrent medications b. Independent Variable: Organ/tissues where tumor recurrence occurred c. Type of Model: Multi-label Classification Model d. Output of Model: Probability of tumor recurrence in each organ/tissue post radiotherapy
15. Radiotherapy Side Effects Grade Model:

(69) A model was built to predict grade of each side effect following radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 15.

(70) TABLE-US-00016 TABLE 15 List of variables considered for modelling Radiotherapy Side effects Grade Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory Body fat disease bowel disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor Tumor volume Tumor vascular Tumor pathology oxygenation density (estimated by imaging) Tumor hydration Tumor marker Tumor marker Tumor marker status (ER) (PR) (HER 2) Tumor marker Ki-67 index Genetic variations Genetic variation (other) irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall hydration status traversed in RT thickness of tissue field Percent of stromal effect tissues proximity necrosis surrounding to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of Temp of organ vs Beam features pre-existing Room free-radical insult at cellular level Total radiation dose per fraction overall treatment planned doses to dose time critical normal tissues site-specific Physician- use of use of hormone patient- reported chemotherapy therapy reported toxicity toxicity use of surgery concurrent medications b. Independent Variable: Grade of Side effect c. Type of Model: Multioutput-Multi-class Classification Model d. Output of Model: Probability of a particular grade of side effect occurring post radiotherapy
16. Radiotherapy Side Effects Time to Resolution:

(71) A model was built to determine time to resolution of a particular side effect post radiotherapy.

(72) This model upon training was used to predict time to resolution for those side effects that were predicted to occur by the Radiotherapy Side effects Model. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 16.

(73) TABLE-US-00017 TABLE 16 List of variables considered for modelling Radiotherapy Side effects Time to Resolution Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g. breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory Body fat disease bowel disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor Tumor volume Tumor vascular Tumor pathology density oxygenation (estimated by imaging) Tumor Tumor marker Tumor marker Tumor marker hydration (ER) (PR) (HER 2) status Tumor Ki-67 index Genetic variations Genetic variation marker irs1 (XRCC2) irs2 (XRCC8) (other) irs3 irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) (RAD51C) XR-1 Distance of organ Vessel wall hydration status (XRCC4) traversed in RT thickness of tissue field Percent of stromal effect tissues proximity necrosis surrounding to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ vs Beam features existing free- Room radical insult at cellular level Total dose per fraction overall treatment planned doses to radiation time critical normal dose tissues site-specific Physician-reported use of use of hormone patient- toxicity chemotherapy therapy reported toxicity use of concurrent surgery medications b. Independent Variable: Time to resolution of side effect c. Type of Model: Survival (Time to event) Model d. Output of Model: Probability of resolution of side effect at time point t (t is a time point like 1 month after radiotherapy)
17. Radiotherapy Non-Resolving Side Effects Model:

(74) A model was built to predict side effects that will not resolve following radiotherapy. a. Dependent Variables: Including but not limited to the list of characteristics presented in Table 17.

(75) TABLE-US-00018 TABLE 17 List of variables considered for modelling Radiotherapy Non- resolving Side Effects Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory Body fat disease bowel disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor Tumor volume Tumor vascular Tumor pathology density oxygenation (estimated by imaging) Tumor Tumor marker Tumor marker Tumor marker hydration (ER) (PR) (HER 2) status Tumor marker Ki-67 index Genetic variations Genetic variation (other) irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall hydration status traversed in RT field thickness of tissue Percent of stromal effect tissues proximity necrosis surrounding to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ vs Beam features existing Room free-radical insult at cellular level Total radiation dose per fraction overall treatment planned doses to dose time critical normal tissues site-specific Physician- use of use of hormone patient- reported chemotherapy therapy reported toxicity toxicity use of surgery concurrent medications b. Independent Variable: Resolution of Side effect c. Type of Model: Multilabel Classification Model d. Output of Model: Probability of resolution of side effects occurring post radiotherapy

EXAMPLES

(76) A] Exemplary System of Determining Tumor Regression, Survival Time and Symptom Improvement Following Radiotherapy

(77) The system receives as input in various embodiments of the claimed invention data including but not limited to the demographic, clinical, social, genomic, omics data and treatment data about the patient as presented in Table 18.

(78) TABLE-US-00019 TABLE 18 List of variables considered for determining tumor regression, survival time and symptom improvement following radiotherapy Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory Body fat disease bowel disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor Tumor volume Tumor vascular Tumor pathology density oxygenation (estimated by imaging) Tumor hydration Tumor marker Tumor marker Tumor marker status (ER) (PR) (HER 2) Tumor marker Ki-67 index Genetic variations Genetic variation (other) irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall hydration status traversed in RT thickness of tissue field Percent of stromal effect tissues proximity necrosis surrounding to critical the tumor structures Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ Beam features existing free- vs Room radical insult at cellular level Total radiation dose per fraction overall treatment planned doses to dose time critical normal tissues site-specific Physician-reported use of use of hormone patient-reported toxicity chemotherapy therapy toxicity use of surgery concurrent medications

Example 1: Survival Prediction Following Radiotherapy 1Adjuvant Radiotherapy

(79) The system consumes the input data and use advanced machine learning and statistical techniques to output a survival score (a value between 0 and 1) which represents the survival likelihood at time t (for example 1 month) following adjuvant radiotherapy.

(80) Using the score, clinicians can intervene and alter the modifiable treatment, patient and clinical characteristics so as to obtain longest survival time (as per progression free survival and overall survival times).

Example 2: Survival Prediction Following Radiotherapy 2Neoadjuvant Radiotherapy

(81) The system consumes the input data and use advanced machine learning and statistical techniques to output a survival score (a value between 0 and 1) which represents the survival likelihood, tumor regression likelihood and symptom improvement likelihood at time t (for example 1 month) following neoadjuvant radiotherapy.

(82) Using the score, clinicians can intervene and alter the modifiable treatment, patient and clinical characteristics so as to obtain maximum tumor regression (as per RECIST criteria), longest survival time (as per progression free survival time) and maximum symptom improvement.

Example 3: Survival Prediction Following Radiotherapy 3Radical Radiotherapy

(83) The system consumes the input data and use advanced machine learning and statistical techniques to output a survival score (a value between 0 and 1) which represents the survival likelihood, tumor regression likelihood and symptom improvement likelihood at time t (for example 1 month) following radical radiotherapy.

(84) Using the score, clinicians can intervene and alter the modifiable treatment, patient and clinical characteristics so as to obtain maximum tumor regression (as per RECIST criteria), longest survival time (as per progression free survival and overall survival times) and maximum symptom improvement.

Example 4: Survival Prediction Following Radiotherapy 4Palliative Radiotherapy

(85) The system consumes the input data and use advanced machine learning and statistical techniques to output a survival score (a value between 0 and 1) which represents the survival likelihood, tumor regression likelihood and symptom improvement likelihood at time t (for example 1 month) following palliative radiotherapy.

(86) Using the score, clinicians can intervene and alter the modifiable treatment, patient and clinical characteristics so as to obtain maximum tumor regression (as per RECIST criteria), longest survival time (as per progression free survival and overall survival times) and maximum symptom improvement.

(87) B] Exemplary System of Determining Side Effects Following Radiotherapy

(88) The system receives as input in various embodiments of the claimed invention data including but not limited to the demographic, clinical, social, genomic, omics data and treatment data about the patient as presented as Table 19.

(89) TABLE-US-00020 TABLE 19 List of variables considered for determining side effects following radiotherapy Age Smoking history Alcohol use Ethnicity Weight Height Volume of Diabetes organ/tissue e.g breast volume in case of breast cancer Hypertension Collagen vascular Inflammatory Body fat disease bowel disease Nutrition Tumor type Tumor stage (TNM) Tumor grade Tumor Tumor volume Tumor vascular Tumor pathology density oxygenation (estimated by imaging) Tumor Tumor marker Tumor marker Tumor marker hydration status (ER) (PR) (HER 2) Tumor marker Ki-67 index Genetic variations Genetic variation (other) irs1 (XRCC2) irs2 (XRCC8) irs3 (RAD51C) irs20 (PRKDC) IRS1-SF (XRCC3) xrs5 (XRCC5) XR-1 (XRCC4) Distance of organ Vessel wall hydration status traversed in RT thickness of tissue field Percent of stromal effect tissues proximity to necrosis surrounding critical structures the tumor Lipid content Bone density Muscle mass hypoxic tissue estimated pH possibility of pre- Temp of organ Beam features existing free-radical vs Room insult at cellular level Total radiation dose per fraction overall treatment planned doses to dose time critical normal tissues site-specific Physician-reported use of use of hormone patient-reported toxicity chemotherapy therapy toxicity use of surgery concurrent medications

Example 5: Adverse Event Prediction Following Radiotherapy

(90) The system consumes the input data and use advanced machine learning and statistical techniques to output an adverse event score (a value between 0 and 1) which represents the adverse events predicted, grade of adverse events predicted, the number of adverse events predicted, the time to resolution of adverse events (for example 1 month) following radiotherapy and non-resolving adverse events.

(91) Using the score, clinicians can intervene and alter the modifiable treatment, patient and clinical characteristics so as to obtain minimum adverse events in terms of number, grade, fastest time to resolution and minimum non-resolving adverse events.

(92) Algorithms:

(93) Survival Scores:

(94) 1. Adjuvant Radiotherapy Survival Score:

(95) Survival Score

(96) Raw Survival Score: Probability of survival at time t

(97) Survival Score=normalize raw survival score to range 0.00 to 1.00.

Example

(98) If Probability of survival at 2 year=0.800 Raw Survival score at time 2 years=0.800 If Highest raw score in training data of probability of survival at time 2 years=0.90 If Lowest raw score in training data of probability of survival at time 2 years=0.15 Survival Score at 2 years=(0.80-0.15)/(0.90-0.15)=0.65/0.75=0.87
Determine Survival Scores for Progression free survival Score (PFS) (primary indicator for success of Adjuvant radiotherapy) and Overall Survival Score (OS) (secondary indicator for success of Adjuvant radiotherapy).
The output of the Radiotherapy Areas of Recurrence Model is the probability of recurrence of the tumor in different organs.
If none of the areas/organs have a probability of recurrence >0.50 then Survival Score (Rec)=0.10
If 1 area/organ have a probability of recurrence >0.50 then Survival Score (Rec)=0.05
If 2 or more areas/organs have a probability of recurrence >0.50 then Survival Score (Rec)=0.0
Survival Score (Adjuvant radiotherapy)=(0.60) (Survival Score (PFS)+(0.30) (Survival Score (OS)+Survival Score (Rec)
Survival Score Interpretation: The higher the patient's survival score, the better the prognosis.
2. Neoadjuvant Radiotherapy Survival Score:
Survival Score

(99) The output of Neoadjuvant Radiotherapy RECIST criteria-based Probability of tumor regression Model is the probability of tumor regression post neoadjuvant radiotherapy. Survival Score (RECIST)=normalize raw survival score to range between 0.00 to 1.00

Example

(100) If Survival probability (RECIST)=0.800 Raw Survival score (RECIST)=0.800 If Highest raw score in training data of probability of tumor regression=0.90 If Lowest raw score in training data of probability of tumor regression=0.15 Survival Score (RECIST)=(0.80-0.15)/(0.90-0.15)=0.65/0.75=0.87 Raw Survival Score (Progression free survival): Probability of survival at time t Survival Score=normalize raw survival score to range 0.00 to 1.00.

Example

(101) If Probability of survival at 2 year=0.800 Raw Survival score at time 2 years=0.800 If Highest raw score in training data of probability of survival at time 2 years=0.90 If Lowest raw score in training data of probability of survival at time 2 years=0.15 Survival Score at 2 years=(0.80-0.15)/(0.90-0.15)=0.65/0.75=0.87
Determine Survival Scores for Progression free survival Score (PFS) (secondary indicator for success of Neoadjuvant radiotherapy)
The output of Neoadjuvant Radiotherapy Symptom Improvement Model is the probability of symptom improvement post neoadjuvant radiotherapy.
Survival Score (SI)=normalize raw survival score to range between 0.00 to 1.00

Example

(102) If Survival probability (SI)=0.800 Raw Survival score (SI)=0.800 If Highest raw score in training data of probability of symptom improvement=0.90 If Lowest raw score in training data of probability of symptom improvement=0.15 Survival Score (SI)=(0.80-0.15)/(0.90-0.15)=0.65/0.75=0.87

(103) The output of the Radiotherapy Areas of Recurrence Model is the probability of recurrence of the tumor in different organs.

(104) If none of the areas/organs have a probability of recurrence >0.50 then Survival Score (Rec)=0.10

(105) If 1 area/organ have a probability of recurrence >0.50 then Survival Score (Rec)=0.05

(106) If 2 or more areas/organs have a probability of recurrence >0.50 then Survival Score (Rec)=0.0

(107) Survival Score (Neoadjuvant radiotherapy)=(0.50) (Survival Score (RECIST))+(0.25)

(108) (Survival Score (PFS)+(0.15) (Survival Score (SI)+Survival Score (Rec)

(109) Survival Score Interpretation: The higher the patient's survival score, the better the prognosis.

(110) 3. Radical radiotherapy Survival Score:

(111) Survival Score Raw Survival Score: Probability of survival at time t Survival Score=normalize raw survival score to range 0.00 to 1.00.

Example

(112) If Probability of survival at 2 year=0.800 Raw Survival score at time 2 years=0.800 If Highest raw score in training data of probability of survival at time 2 years=0.90 If Lowest raw score in training data of probability of survival at time 2 years=0.15 Survival Score at 2 years=(0.800.15)/(0.900.15)=0.65/0.75=0.87
Determine Survival Scores for Progression free Survival Score (PFS) and Overall Survival Score (OS).
The output of Radical Radiotherapy RECIST criteria-based Probability of tumor regression Model is the probability of tumor regression post radical radiotherapy.
Survival Score (RECIST)=normalize raw survival score to range between 0.00 to 1.00

Example

(113) If Survival probability (RECIST)=0.800 Raw Survival score (RECIST)=0.800 If Highest raw score in training data of probability of tumor regression=0.90 If Lowest raw score in training data of probability of tumor regression=0.15 Survival Score (RECIST)=(0.800.15)/(0.900.15)=0.65/0.75=0.87

(114) The output of Radical Radiotherapy Symptom Improvement Model is the probability of symptom improvement post neoadjuvant radiotherapy.

(115) Survival Score (SI)=normalize raw survival score to range between 0.00 to 1.00

Example

(116) If Survival probability (SI)=0.800 Raw Survival score (SI)=0.800 If Highest raw score in training data of probability of symptom improvement=0.90 If Lowest raw score in training data of probability of symptom improvement=0.15 Survival Score (SI)=(0.800.15)/(0.900.15)=0.65/0.75=0.87

(117) The output of the Radiotherapy Areas of Recurrence Model is the probability of recurrence of the tumor in different organs. If none of the areas/organs have a probability of recurrence >0.50 then Survival Score (Rec)=0.10 If 1 area/organ have a probability of recurrence >0.50 then Survival Score (Rec)=0.05 If 2 or more areas/organs have a probability of recurrence >0.50 then Survival Score (Rec)=0.0 Survival Score (Radical radiotherapy)=(0.50) (Survival Score (OS)+(0.20) Survival Score (RECIST)+(0.125) (Survival Score (PFS))+(0.075) (Survival Score (SI)+Survival Score (Rec) Survival Score Interpretation: The higher the patient's survival score, the better the prognosis.
4. Palliative Radiotherapy Survival Score:

(118) The output of Palliative Radiotherapy Symptom Improvement Model is the probability of symptom improvement post Palliative radiotherapy.

(119) Survival Score (SI)=normalize raw survival score to range between 0.00 to 1.00

Example

(120) If Survival probability (SI)=0.800 Raw Survival score (SI)=0.800 If Highest raw score in training data of probability of symptom improvement=0.90 If Lowest raw score in training data of probability of symptom improvement=0.15 Survival Score (SI)=(0.800.15)/(0.900.15)=0.65/0.75=0.87

(121) The output of Palliative Radiotherapy RECIST criteria-based Probability of tumor regression Model is the probability of tumor regression post palliative radiotherapy. Survival Score (RECIST)=normalize raw survival score to range between 0.00 to 1.00

Example

(122) If Survival probability (RECIST)=0.800 Raw Survival score (RECIST)=0.800 If Highest raw score in training data of probability of tumor regression=0.90 If Lowest raw score in training data of probability of tumor regression=0.15 Survival Score (RECIST)=(0.800.15)/(0.900.15)=0.65/0.75=0.87 Survival Score

(123) Raw Survival Score: Probability of survival at time t

(124) Survival Score=normalize raw survival score to range 0.00 to 1.00.

Example

(125) If Probability of survival at 2 year=0.800 Raw Survival score at time 2 years=0.800 If Highest raw score in training data of probability of survival at time 2 years=0.90 If Lowest raw score in training data of probability of survival at time 2 years=0.15 Survival Score at 2 years=(0.800.15)/(0.900.15)=0.65/0.75=0.87
Determine Survival Scores for Progression free survival Score (PFS) and Overall Survival Score (OS).
The output of the Radiotherapy Areas of Recurrence Model is the probability of recurrence of the tumor in different organs.
If none of the areas/organs have a probability of recurrence >0.50 then Survival Score (Rec)=0.10
If 1 area/organ have a probability of recurrence >0.50 then Survival Score (Rec)=0.05
If 2 or more areas/organs have a probability of recurrence >0.50 then Survival Score (Rec)=0.0
Survival Score (Palliative Radiotherapy)=(0.50) (Survival Score (SI)+(0.20) (Survival Score (RECIST)+(0.125) (Survival Score (PFS))+(0.075) (Survival Score (OS))+Survival Score (Rec)
Survival Score Interpretation: The higher the patient's survival score, the better the prognosis.
Adverse Event Score:

(126) The output of the Radiotherapy side effects grade model is the probability of different side effects and the grade of each If highest AE grade is Grade 5 (death).fwdarw.then AE Score=1.0 If highest AE grade is Grade 4 (Life threatening).fwdarw.then AE score=0.8 Else Algorithm to calculate AE score is as follows: Take grades of 5 predicted adverse events with highest grade
Multiply probability of each adverse event with 0.15 if it is grade 3, 0.1 if grade 2 and 0.05 if grade 1.
AE (Grade)=Sum of 5 products described above

Example

(127) Nausea Grade 3 with probability 0.8=0.150.8=0.12 Constipation Grade 2 with probability 0.5=0.10.5=0.05 Cough Grade 1 with probability 0.9=0.050.9=0.045 No other side effects AE (Grade)=0.12+0.05+0.045+0+0=0.215
The output of Radiotherapy Side effects time to resolution model is the probability of a side effect resolving at time t
Raw Score: Probability of AE resolution at time t
AE resolution Score=normalize raw score to range 0.00 to 1.00.

Example

(128) If Probability of AE resolution at 1 month=0.800 Raw AE resolution score at time 1 month=0.800 If Highest raw score in training data of probability of AE resolution at time 1 month=0.90 If Lowest raw score in training data of probability of AE resolution at time 1 month=0.15 Raw AE resolution Score at 1 month=(0.800.15)/(0.900.15)=0.65/0.75=0.87 AE (Res)=1(the average of 5 highest AE resolution scores at 1 month)
The output of the Radiotherapy non-resolving side effects model is probability of resolution of side effects post radiotherapy.
If all of the AE have a probability of resolution >0.50 then AE (Non-Resolution)=0.00
If 1 SE has a probability of resolution <0.50 then AE (Non-Resolution)=0.05
If 2 or more SE have a probability of resolution <0.50 then AE (Non-Resolution)=0.10
AE Score=AE (grade)+(0.15) AE (Res)+AE (Non-Resolution)
AE Score Interpretation: The Lower the AE Score, the Better the Quality of Life (QoL) of the Patient

(129) Once the data is entered into the system, then the process of score generation shall initiate. The basic steps involved include the following

(130) All the details shall be entered into the respective fields. The values are then classified and given a unique value based on the relative importance of the same. Weighted averages from literature/past experience from center and feedback loop from clinician/radiation oncologist were considered as next level of inputs, which shall be going through the ANN (artificial Neural network system).

(131) While considering all the above parameters as mentioned in the embodiment, the following assumptions are made: a. Artificial neural network (ANN) using the clinical data including patient visit data, admission data, adverse event data, emergency department visit data, family history, medical history, survival data, treatment data and treatment response data and genomic data including gene expression and gene mutation data, radiation planning details, beam strength, contour and other shall go as inputs. The trained ANN model will predict the response rates/relapse probability and survival probability at time t for the patient. a. Survival Score b. Raw Survival Score: Probability of survival at time t c. Survival Score=normalize raw survival score to range 0.00 to 1.00. d. Similarly, the probability of response is assessed in terms of i) Magnitude of response (regression from baseline) ii) Time for maximum response iii) Score at various permutations (exif we increase dosethen what shall be response, if we increase oxygen concentration, if we give more fractions etc.) iv) Durability of response v) Probability of response failure vi) And other desired variables
Similarly, the probability of response is assessed in terms of i) Probability of normal organ damage ii) Time for maximum toxicity/highest grade (AE) iii) Intensity of maximum toxicity (grade of AE) iv) Duration of toxicity v) Time for resolution vi) Probability of residual/long term toxicity vii) Score at various permutations (exif we reduce dosethen to what extent we can reduce the toxicity etc., what duration of break shall benefit the recovery of organ)

(132) The System gives a customized survival prediction and AE prediction score to each patient taking into account lot of the clinical and genomic data of the patients including treatment data.