NOVEL NANOTECHNOLOGY-DRIVEN PROTOTYPES FOR AI-ENRICHED BIOCOMPATIBLE PROSTHETICS FOLLOWING EITHER RISK OF ORGAN FAILURE OR MODERATE TO SEVERE IMPAIRMENT

Abstract

Three groups of biocompatible implants were created, to leverage physiological impairment caused by (i) cardiovascular, (ii) renal, and (iii) neuronal diseases. Each group of implants is subdivided into three categories according to extra functionality added plus integrated additions. The first generation contains basic functionality and the second and third generations contain extra functions. Finally, further additions can be combined and integrated. Therefore, the first group comprises of the “First Generation of Cardiovascular Implants” plus the “Second Generation of Cardiovascular Implants” plus the “Third Generation of Cardiovascular Implants” plus additional integrations named “Additions”. Equally, the second group comprises of the “First”, the “Second” and the “Third” Generation of Renal Prosthetics plus Additions. The same categorisation applies to Neural Implants, which are three generations plus additions. This can be found in the description of claims presented in the Austrian Prio (provisional patent application) number A 60273/2019, from 11 Dec. 2019.

Claims

1. Heart Implant (1st generation): A soft biocompatible membrane mimicking the heart anatomy, which the inventor here calls biocompatible matrix or “the shell”. Manufacturing, biocompatible materials and microchips are indicated in the Description. This shell contains microprobes and electrodes to work as conventional heart implants—e.g., pacemakers, CDIs, and resynchronisers. This shell applies mechanical forces on the heart in cases of severe heart electric failure (resuscitation). This shell is coupled with sensors for monitoring vital signals and with microchips-based actuators that acts on the heart. This shell is coupled with an AI-driven microcontroller that receives signals from sensors, analyses these signals, sends action signals to microchips-based actuators, and coordinates/controls the whole system. The shell, in severe cases of mechanical impairment affecting heart rate, is either coupled or replaced by an additional inner membrane (shell) placed to mechanically expand and contract inside the heart chambers, assuring that a reasonable level of blood flow rate is preserved. This inner shell also corrects the heart valves' movement, if needed. Software for AI-driven implant control is installed in a device similar to a pacemaker, which is superficially implanted under the skin, as usual—AI microcontroller).

2. The shell according to claim 1 has a flexible design and can be built in parts—e.g. (1) only resynchronisation and implantable cardioverter-defibrillator sensors and microchips/electrodes placed; (2) only pacemaker placed; (3) the whole composite plus mechanical compressor plus drug-delivery nano-complexes placed. Different from conventional pacemakers, CDIs, and resynchronisers, the shell possesses all these functionalities plus the application of mechanical forces whether electrical impulses fade. This shell is also structurally distinct from conventional heart implants, because (i) it is a shell that covers the outer cardiac structure and (ii) a second shell layer can be placed in the inner heart to improve aid.

3. Heart Implant (2nd generation): A denser matrix (3D structural construction using silicone elastomers and polymers). There are three possible configurations for this shell, which apply according to disease severity and clinical indication: (i) external and covering the organ to compress and distend, according to claim 1; (ii) inflated blocks within heart cavities to expand and contract, according to claim 2; or (iii) disconnected from the heart cavity and functioning as an implanted ventricular assist device (VAD), whose dynamic functionality is provided by AI-guided sensors and microchips, and can endure a longer lifespan.

4. Heart Implant (3rd generation): Either a shell according to claim 1 or a denser matrix VAD according to claim 3 wherein contains additional structures for therapeutic nano composites, delivery systems, and imaging. Example: to control the injection of stem cells-based therapeutics and to deliver medicinal compounds locally, to restore damaged tissue and local signalling cascades. These nano-composites are attached to the implant, at locations prescribed by clinicians, which depends on each patient's case.

5. Renal Prosthetics (1st generation): Set of sensors and micro actuators that are connected to a control system (microchip). Biomaterials and biocompatible microchips are indicated in the Description. Microelectronics (sensors) are used to capture physiological signals and to take measurements (e.g., real time dosage of creatinine in both urine and blood, real time dosage of acid uric in both blood and urine, monitoring of inflammatory biomarkers, and calculation of nephrons' filtering capacity). The AI-control system deployed in a microchip (microcontroller) forecasts renal failure using these variables and responds to imminent threats by sending signals to the actuators, to maintain local homeostasis at acceptable levels. This implant via AI-control coordinates (i) real-time renal function monitoring, (ii) drug delivery, (iii) imaging, and (iii) regenerative tissue therapeutics (e.g., based on stem cells technology) in patient with reduced renal capacity, without the indication of nephrectomy. The diseased organ is constantly monitored and treated.

6. Renal Prosthetics (2nd and 3rd generations): 3D printed reconstructed structure replicating the patient kidney's target volume, whose design is personalised, according to 3D reconstructed and segmented CT/MRI data, without following anatomic patterns, precisely. This structure contains multiple chambers, filters, and valves to filter the blood and pump both clean blood and residual fluid (artificial urine) using a target chamber volume. To adopt the system with important functional capabilities, if surrounding impaired tissue or local homeostasis need to be treated, therapeutic components (e.g., drug delivery and stem cells technology) are used. Sensors are strategically placed to monitor the prosthetics functionality for mitigating faults, measure physiological flow drivers (e.g., gradient of pressure) and concentration of blood compounds (e.g., concentration of dialysed uric acid) to control filtering and pumping mechanisms, along with concentration of chemical compounds leading to physiological impairment such as hypocalcaemia, to feedback safety and alert mechanisms.

7. Neural Implant (1st generation): Nano-composites combined with sensors and actuators that are controlled by AI technology, for signals analysis, dynamically monitoring and triggering the delivery of chemicals to brain tissue. Three variations of the model are presented. They have a similar structure, varying only in target signalling cascade (or group of neuronal cells to be treated) and needed compound to be delivered. The materials used to encapsulate implanted devices and to form in artificial shells (artificial tissues) are indicated in the Description. Microchips used in sensors for electrical signals registration probes, in imaging, and in AI-based implant's control are also indicated in the Description. These nano-composites combined with sensors, actuators and an AI-control system monitor, analyse and control/adjust chemical reactions and the relevant signalling cascades. The variations of the model are as follows. Model 1 is a bio-implant that collects local electrical and biochemical signals and use AI technology to drive immune assays for biomarker determination and knockdown of diseased signalling networks. The target is amyloid and/or tau, modulating diseased signalling cascades, to mitigate Alzheimer plaque build-up. Model 2 is a bio-implant for local physiological monitoring, dynamically delivering compounds, to mitigate excitotoxicity caused by imbalance in expression of neurotransmitters. Model 3 is a bio-implant for drug delivery and disease progression follow-up (dynamically capturing signals during treatment, which are analysed in real time, to monitor progress and trigger local drug administration accordingly, in response to different physiological responses). This model is used for tissue recovery, following for instance brain injury caused by strokes or cranial traumatism resulting from road accidents, and other critical episodes.

8. Neural Implant (2nd generation): A set of signals transmitters leveraging electromagnetic dysfunction. The transmitters communicate with an AI platform for signals processing and analysis. Again, the materials used to encapsulate implanted devices and to form artificial shells (artificial tissues), microchips used in sensors for electrical signals registration probes, in imaging, and in AI-based implant's control are indicated in the Description. The transmitters coupled with AI microcontrollers monitor, analyse and control/adjust electromagnetic signals that are related to functional impairment like visual dysfunction and hearing loss. There are two variations of the model, as follows. Model 1 is a biocompatible electrical encoder-decoder implant that supports the transmission of visual information from the retina to the brain, when the process is compromised by optical nerve damage, mitigating vision loss (e.g., in patients suffering from glaucoma). Model 2 is a biocompatible electrical encoder-decoder implant that supports the transmission of sound from the cochlea to the brain, when the process is compromised by auditory nerve fibre damage, mitigating hearing loss.

9. In all the implants here presented, a complete AI platform was developed to control the system, to design the implants in a personalised manner based on patients' CT/MRI images, and to generate updates for the algorithms deployed on the microcontrollers (i.e., the algorithms that analyses signals and control the implants). This AI platform also contains a virtual environment to plan computer-guided robotic surgery. The signals analysed are collected into variables. These variables are divided into four main groups according to their usage (indicated in the Description).

10. As an additional feature, the implants send recorded signals to (i) a light-based alert microchip implanted in the patient's wrist, (ii) a computer located in the hospital where the patient is treated, and (iii) the patient's mobile device. This is for safety and to store signals for software updates.

Description

BRIEF DESCRIPTION OF DRAWINGS

[0076] FIG. 1. Illustration of shell design. The first step in the manufacture process is to generate the structural models, which can be done via most of the existing segmentation and 3D reconstruction software. The author here indicates the use of Materialise 3-matic, version 13.0. In the diagram, one can see (a) how CT/MRI images are used for matching patients' anatomy and the implanted membrane; (b) how the images are automatically segmented for extracting the regions of interest (ROI); (c) and (d) how remaining artefacts are manually removed and the resulting structure prepared for further filtering; (e), (0, and (g) the final coupling membrane, the removal of the inner part of the solid, the shell thickness being fixed at 1-2 mm, and the final structure further segmented according to the regions of interest; and (h) an illustration of the diverse structural parts that can be manufactured, on the basis of anatomic CT/MRI images. Further filtering can be employed to smooth the superficial layer, removing final artefacts. The final anatomic models are saved in stl format for further use.

[0077] FIG. 2. Illustration of the coupling shell-organ. Diagrams a and b illustrate normal and impaired diastole-systole, respectively. Indeed, illustrations a and b left indicate heart relaxation, while illustrations a and b right illustrate heart muscular contraction. In diagram b right, it can be seen that systole is compromised by reduced ability of the muscle fibres to contract. Diagram c illustrates restoring contraction potential resulting from mechanically driven implant aid. Diagram c also shows a close-up view of the coupling shell-tissue. Illustration d presents the design of the virtual environment to be used by medical experts, in order to perform shell-organ coupling and robotics driven protocol generation, preceding a computer-assisted surgery.

[0078] FIG. 3. Illustration of prosthetics implantation. The proposed implanted modulus are shown. Pacemakers, resynchronisers, and CDIs are introduced intra-venous, being implanted in the cardiac muscle, as usual. The mechanical pacemaker here presented (mechanical compression in case of severe electrical failure) is designed to form a shell (soft matrix) and is introduced externally, to apply contractile forces on the heart chambers. However, in severe cases, where myocardial stiffness increases abnormally and the cardiac valves become defective, implanted modules are introduced within the heart chambers, alternatively, either inflated within the chambers (flexible silicone-based membranes) functioning as inner flexible pumps or fixed to the heart valves to facilitate their movement.

[0079] FIG. 4. Schematic view of the prosthetic components and operational modes: (a) illustration of groups of components; (b) operational modes; and (c) security and functional management. This figure shows a simplified overview of varied operational modes found in the model: (i) pacemaker, (ii) defibrillator, and (iii) mechanical compressor. These operational modes rely on both physical components and mathematical abstraction translated into algorithms implemented via computational pieces of software. Regarding physical components, three major groups of components were designed: (i) sensors, (ii) actuators, and (iii) optional components as nano-complexes for imaging, drug delivery, and stem cells therapy. The group of optional components are classified as biodegradable and resident, according to the respective medical prescription. Finally, three components are used for safety and functional management. These are a programmed micro-chip (AI microcontroller) that performs AI-driven decision making; a micro light-monitor that alerts on critical conditions using green, yellow, and red lights; and a computational platform that is placed remotely (e.g., in hospitals and clinics) operating as a General Control Prosthetics (GCP) platform. The GCP platform also hosts an advanced version of the AI-model named backtrack-convnet, which is coupled with a reinforcement learning algorithm, in an adversarial network architecture, which receives signals from the implanted sensors and dynamically interacts with the backtrack-convnet, to update it dynamically, giving raise to its new versions, in order to periodically update the AI microcontroller.

[0080] FIG. 5. Illustration of functional abstraction (data transmission and GCP platform): (a) sensors transmit signals to the AI microcontroller, which are analysed and translated into commands that are sent to the actuators; (b) signals are also sent by the sensors to both the light monitor for security alerts and to the GCP platform for clinical monitoring and AI model improvement, using newly collected data; and (c) clinical monitoring is performed remotely, on the GCP platform, and in the same monitoring system, generation of AI model updates takes place, using a newly created backtrack-convnet in an adversarial network architecture.

[0081] FIG. 6. Schematic view of the AI model architecture. The AI model comprises of a robust convolutional neural network with batch normalisation and exponential linear units, using a cropped training strategy implemented in python (Anaconda). It was trained on the basis of thousands of historical records and synthetic data was employed to assess the bias-variance trade-off. Multi-classification on tuples (input signals, target action mode) was performed using a new subset of the original dataset (the validation set).

[0082] FIG. 7. Results showing ConvNet's performance selecting action mode during myocardial dysfunction. Quality assurance is indicated in diagrams (a) and (b), which show the learning curves as a function of the epochs and batches, along with sampling size, respectively. Diagram (c) shows the model generalisation power and diagram (d) illustrates model application. Given the results, high volumes of data can be collected over a reasonable period—i.e. about 13 days, assuming that measurements follow the physiological interval between consecutive heartbeats, which is 1.1 seconds in average; in candidates for receiving these implants. This would empower the AI microcontroller with dynamic self-calibration, resulting in model re-training to assure consistency and dynamic self-adaptation to changing conditions (sampling variance), following surgery. Therefore, the initial trained model can adapt to changing conditions, on the basis of continuous data collection and monitoring.

[0083] FIG. 8. Illustration of shell design showing complete kidney replacement: (a) and (b) CT scan segments; (c) 3D reconstruction of left and right kidneys; (d) left and rights kidneys' volume extraction and imprint of sensors placement, to be followed in the artificial 3D printed prototype.

[0084] FIG. 9. Graphical summary of the kidney morphology, which defines the characteristics of the prosthetics; and illustration of sensors placement, which correlates with imprints determined via 3D anatomical reconstruction.

[0085] FIG. 10. Schematic view of the artificial chamber designed to filter blood. Diagram (a) shows the major chamber parts—the shell, the renal artery connection valve, the upper and the lower waste cavities, the middle clean blood cavity, the ureters connection valves, the renal vein connection valve, the volume markers and the filters simulating the three main glomerular capillary structures involved in blood filtering (the endothelial pores of 70 to 100 nanometres in diameter, the basement membrane region, and the epithelial podocytes area). Diagrams (b) and (c) indicate the envisioned haemodialysis cycle, which comprises of 8 steps. First, renal artery connector opens. Second, the chamber fills in. Third, renal artery connector closes. Fourth, middle particles (predominantly) filtering occur in the upper waste cavity. Fifth, essential particles remain in the middle clean blood cavity and small particles are collected in the lower waste cavity. Sixty, the volume markers close all cavities. Seventh, clean blood in the middle cavity is pumped through the renal vein connection and waste solution (artificial urine) stored in both the upper and the lower cavities are pumped through the ureters' connections. Eighth, all the volume markers open, finishing the cycle, allowing the renal artery connection to open, once more, and the process to continue.

[0086] FIG. 11. Results showing Gboost-based model's performance managing prosthetics operation, on the basis of filtration time as a function of target molecules concentration and physiological condition. Quality assurance is indicated in diagrams (a) and (b), which show comparison with other strategies (ROC curves) and the error (3-fold avegare) vs. epochs for early stop, respectively. Diagram (c) shows model performance as a function of sampling size.

BIBLIOGRAPHY

[0087] Moglia A, Menciassi A, Dario P, Cuschieri A (2009) Capsule endoscopy: Progress update and challenges ahead. Nat. Rev. Gastroenterol. Hepatol. 6:353-361 [0088] Canales A, Jia X, Froriep U P, et al (2015) Multifunctional fibers for simultaneous optical, electrical and chemical interrogation of neural circuits in vivo. Nat Biotechnol 33:277-284. doi: 10.1038/nbt.3093 [0089] Pereira G (2017a) Genomics and Artificial Intelligence Working Together in Drug Discovery and Repositioning: The Advent of Adaptive Pharmacogenomics in Glioblastoma and Chronic Arterial Inflammation Therapies. In: Malik S (ed) Biotechnology and Production of Anti-Cancer Compounds. Springer-Nature, p 30 [0090] Pereira G C (2019) Application of Biotechnology in Producing Plant Bio-active Compounds. In: Natural Bio-active Compounds. Springer Singapore, pp 59-78 [0091] Pereira G C, Malik S, Kis Z, Rocamonde B (2019) Computationally Designed Recombinant-DNABased Compounds Production Driven in Plants During Secondary Metabolism and Their Implication in Antimalarial Therapies. In: Natural Bio-active Compounds. Springer Singapore, Singapore, pp 127-146 [0092] Boutry C M, Beker L, Kaizawa Y, et al (2019) Biodegradable and flexible arterial-pulse sensor for the wireless monitoring of blood flow. Nat Biomed Eng 3:47-57. doi: 10.1038/s41551-018-0336-5 [0093] Shin J, Yan Y, Bai W, et al (2019) Bioresorbable pressure sensors protected with thermally grown silicon dioxide for the monitoring of chronic diseases and healing processes. Nat Biomed Eng 3:37-46. doi: 10.1038/s41551-018-0300-4 [0094] Seignez C, Phillipson M (2017) Implanted biomaterials: Neutrophil-mediated vascularization. Nat. Biomed. Eng. 1 [0095] Bank R A (2019) Limiting biomaterial fibrosis. Nat. Mater. 18:781 Bannerman D, Pascual-Gil S, Radisic M (2019) An optimal gel patch for the injured heart. Nat Biomed Eng 3:592-593. doi: 10.1038/s41551-019-0440-1 [0096] Liu Y, Liu J, Chen S, et al (2019b) Soft and elastic hydrogel-based microelectronics for localized low-voltage neuromodulation. Nat Biomed Eng 3:58-68. doi: 10.1038/s41551-018-0335-6 [0097] Conte R, Marturano V, Peluso G, et al (2017) Recent Advances in Nanoparticle-Mediated Delivery of Anti-Inflammatory Phytocompounds. Int J Mol Sci 18:709. doi: 10.3390/ijms18040709 [0098] Singh M, Devi S, Rana V S, et al (2019) Delivery of phytochemicals by liposome cargos: recent progress, challenges and opportunities. J Microencapsul 36:215-235. doi: 10.1080/02652048.2019.1617361 [0099] Pereira G C (2020c) Governing issues in nanoscale systems and their potential for improving the therapeutic application of phytoconstituents. In: Swamy M K, Saeed M (eds) Plant-derived Bioactives—Chemistry and Mode of Action. Springer Nature Liu T, Liu S, Tu K H, et al (2019a) Crested two-dimensional transistors. Nat. Nanotechnol. 14:223-226 [0100] Masvidal-Codina E, Illa X, Dasilva M, et al (2019) High-resolution mapping of infraslow cortical brain activity enabled by graphene microtransistors. Nat Mater 18:280-288. doi: 10.1038/s41563-018-0249-4 [0101] Tite T, Chiticaru E A, Burns J S, Ionitâ M (2019) Impact of nano-morphology, lattice defects and conductivity on the performance of graphene based electrochemical biosensors. J. Nanobiotechnology 17 [0102] Liu H, Zhao T, Jiang W, et al (2015) Flexible Battery-Less Bioelectronic Implants: Wireless Powering and Manipulation by Near-Infrared Light. Adv Funct Mater 25:7071-7079. doi:10.1002/adfm.201502752 [0103] Anumanchipalli G K, Chartier J, Chang E F (2019) Speech synthesis from neural decoding of spoken sentences. Nature 568:493-498. doi: 10.1038/s41586-019-1119-1 [0104] Akbari H, Khalighinejad B, Herrero J L, et al (2019) Towards reconstructing intelligible speech from the human auditory cortex. Sci Rep 9: doi: 10.1038/s41598-018-37359-z [0105] Gholami B, Haddad W M, Bailey J M (2018) AI in the ICU: In the intensive care unit, artificial intelligence can keep watch, IEEE Spectr 55:31-35. doi:10.1109/MSPEC.2018.8482421 [0106] Pereira G (2020a) Computational Approaches in Drug Development and Phytocompounds Analysis. In: Swamy M K, Saeed M (eds) Plant-derived Bioactives—Chemistry and Mode of Action. Springer Nature [0107] Kumar S N, Lenin Fred A, Ajay Kumar H, et al (2020) Segmentation of Anomalies in Abdomen C T Images by Convolution Neural Network and Classification by Fuzzy Support Vector Machine. pp 157-196 [0108] Nagarajan G, Sathish Kumar B S (2020) Proficient Reconstruction Algorithms for Low-Dose X-Ray Tomography. pp 237-256 [0109] Pereira G C (2017b) Genomics and Artificial Intelligence Working Together in Drug Discovery and Repositioning: The Advent of Adaptive Pharmacogenomics in Glioblastoma and Chronic Arterial Inflammation Therapies. In: Biotechnology and Production of Anti-Cancer Compounds. Springer International Publishing, Cham, pp 253-281 [0110] Pereira G (2020b) Polyphenols role in autoimmune and chronic inflammatory diseases and the advent of computer-driven plant therapies. In: Swamy M K, Saeed M (eds) Plant-derived Bioactives—Chemistry and Mode of Action. Springer Nature [0111] NHSBT (2017) NHSBT Organ Donation and Transplantation: Activity Figures for the UK-2017 statistical report [0112] WHO (2019) WHO|Outcomes of organ transplantation. https://www.whoint/transplantation/gkt/statistics/kidney_outcomes/en/. Accessed 16 Dec. 2019 [0113] NHSBT (2019a)NON-PROCEEDING DECEASED DONORS NHSBT (2019b) Organ Donation and Transplantation Statista (2019) Patient deaths on organ transplant waiting list 2019 Statista. https://www.statista.com/statistics/519829/patient-deaths-on-organ-transplant-waiting-listunited-kingdom-uk/. Accessed 16 Dec. 2019 [0114] NKF (2016) Organ Donation and Transplantation Statistics 1National Kidney Foundation. https://www.kidney.org/news/newsroom/factsheets/Organ-Donation-and-Transplantation-Stats. Accessed 16 Dec. 2019 [0115] Health Resources & Services Administration (2019) Organ Donation Statistics Organ Donor. https://www.organdonor.gov/statistics-stories/statistics.html. Accessed 16 Dec. 2019 [0116] Organ Donor Foundation (2016) Organ Donor Foundation—Statistics. https://www.odf.org.za/infoand-faq-s/statistics.html. Accessed 16 Dec. 2019 [0117] Chan-on C, Sarwal M M (2017) A comprehensive analysis of the current status and unmet needs in kidney transplantation in Southeast Asia. Front. Med. 4 [0118] Stouffer G A (2008) Cardiovascular Hemodynamics for the Clinician. Blackwell Publishing Ltd.