Aspects of the present disclosure relate to systems and methods for manufacturing biologically-produced pharmaceutical products. Some of the systems described herein comprise an upstream component comprising a bioreactor and at least one filter (e.g., a filter probe) integrated with a downstream component comprising a purification module comprising at least a first partitioning unit and a second partitioning unit. In some embodiments; these integrated biomanufacturing systems may be operated under continuous or conditions and may be capable of efficiently producing pure, high-quality pharmaceutical products.

In order to facilitate product development, such as pharmaceutical product development, a computer implemented method and an apparatus are proposed that enable formulators to develop robust drug formulations in a cost- and time-efficient manner. To start the development process, the user selects the preferred dosage form (e.g., granules, pellets, capsules, tablets etc.), defines a target profile (e.g., amount of active ingredient per unit, size of dosage form, mechanical strength of dosage form, desired release behaviour etc.) and enters key characteristics of the active ingredient (e.g., true density, particle size distribution data, bulk and tapped density, angle of repose, compressibility and compactibility profile etc.). The identity (e.g., chemical name or structure) of the active ingredient is not necessarily disclosed. The apparatus processes the provided data and calculates key parameters of the AI (e.g., particle size, powder density, powder flow and tabletability) Similar key parameters are calculated for common pharmaceutical excipients and stored in the apparatus. The apparatus then selects all relevant excipients and suggests a suitable manufacturing process. Combinations of active ingredients and excipients qualify as drug formulation if the predicted properties comply with the defined target profile. The following aspects can be considered: solubility and permeability of the active ingredient, dissolution of the active ingredient, probability to pass the content uniformity criteria, flowability of the powder blend, tabletability of the powder blend, mechanical strength and size of the tablet, compatibility of active ingredients and excipients etc.

In order to facilitate product development, such as pharmaceutical product development, a computer implemented method and an apparatus are proposed that enable formulators to develop robust drug formulations in a cost- and time-efficient manner. To start the development process, the user selects the preferred dosage form (e.g., granules, pellets, capsules, tablets etc.), defines a target profile (e.g., amount of active ingredient per unit, size of dosage form, mechanical strength of dosage form, desired release behaviour etc.) and enters key characteristics of the active ingredient (e.g., true density, particle size distribution data, bulk and tapped density, angle of repose, compressibility and compactibility profile etc.). The identity (e.g., chemical name or structure) of the active ingredient is not necessarily disclosed. The apparatus processes the provided data and calculates key parameters of the AI (e.g., particle size, powder density, powder flow and tabletability) Similar key parameters are calculated for common pharmaceutical excipients and stored in the apparatus. The apparatus then selects all relevant excipients and suggests a suitable manufacturing process. Combinations of active ingredients and excipients qualify as drug formulation if the predicted properties comply with the defined target profile. The following aspects can be considered: solubility and permeability of the active ingredient, dissolution of the active ingredient, probability to pass the content uniformity criteria, flowability of the powder blend, tabletability of the powder blend, mechanical strength and size of the tablet, compatibility of active ingredients and excipients etc.

An isotope ratio spectrometer is operated for measurement of a sample. First isotope ratios and first signal intensities are measured for a reference in the spectrometer, over a first measurement time period. A first relationship comprising a relationship between the first isotope ratios and the first signal intensities is determined. Sample isotope ratios and sample signal intensities are measured in the spectrometer, over a second measurement time period subsequent to the first measurement time period. Second isotope ratios and second signal intensities for a reference are measured in the spectrometer, over a third measurement time period subsequent to the second measurement time period. A second relationship comprising a relationship between the second isotope ratios and the second signal intensities is determined. A reference isotope ratio is estimated for a time X within the second measurement time period, based on the first relationship and the second relationship.

An isotope ratio spectrometer is operated for measurement of a sample. First isotope ratios and first signal intensities are measured for a reference in the spectrometer, over a first measurement time period. A first relationship comprising a relationship between the first isotope ratios and the first signal intensities is determined. Sample isotope ratios and sample signal intensities are measured in the spectrometer, over a second measurement time period subsequent to the first measurement time period. Second isotope ratios and second signal intensities for a reference are measured in the spectrometer, over a third measurement time period subsequent to the second measurement time period. A second relationship comprising a relationship between the second isotope ratios and the second signal intensities is determined. A reference isotope ratio is estimated for a time X within the second measurement time period, based on the first relationship and the second relationship.

A variational autoencoder (VAE) has been developed to learn a continuous numerical, or latent, representation of molecular structure to expand reference libraries for small molecule identification. The VAE has been extended to include a chemical property decoder, trained as a multitask network, to shape the latent representation such that it assembles according to desired chemical properties. The approach is unique in its application to metabolomics and small molecule identification, focused on properties that are obtained from experimental measurements (m/z, CCS) paired with its training paradigm, which involves a cascade of transfer learning iterations. First, molecular representation is learned from a large dataset of structures with m/z labels. Next, in silico property values are used to continue training. Finally, the network is further refined by being trained with the experimental data. The trained network is used to predict chemical properties directly from structure and generate candidate structures with desired chemical properties. The network is extensible to other training data and molecular representations, and for use with other analytical platforms, for both chemical property and feature prediction as well as molecular structure generation.

A variational autoencoder (VAE) has been developed to learn a continuous numerical, or latent, representation of molecular structure to expand reference libraries for small molecule identification. The VAE has been extended to include a chemical property decoder, trained as a multitask network, to shape the latent representation such that it assembles according to desired chemical properties. The approach is unique in its application to metabolomics and small molecule identification, focused on properties that are obtained from experimental measurements (m/z, CCS) paired with its training paradigm, which involves a cascade of transfer learning iterations. First, molecular representation is learned from a large dataset of structures with m/z labels. Next, in silico property values are used to continue training. Finally, the network is further refined by being trained with the experimental data. The trained network is used to predict chemical properties directly from structure and generate candidate structures with desired chemical properties. The network is extensible to other training data and molecular representations, and for use with other analytical platforms, for both chemical property and feature prediction as well as molecular structure generation.

A device for providing drug verification may work in conjunction with drug delivery devices such as medical pumps to provide a chemical and concentration analysis of drugs being delivered forming a signature that can be compared to a signature associated with the proper drug, reducing errors in medicine delivery and ensuring proper use of medicines throughout their lifecycle.

A device for providing drug verification may work in conjunction with drug delivery devices such as medical pumps to provide a chemical and concentration analysis of drugs being delivered forming a signature that can be compared to a signature associated with the proper drug, reducing errors in medicine delivery and ensuring proper use of medicines throughout their lifecycle.

Computer-based methods that permit two or more users to perform simultaneous edits on a digitally encoded molecular structure. The methods use properties of conflict-free replicated data types (CRDT's) and causal trees to provide a distributed system which can manage the life-cycle of virtual molecular structures; including simultaneous editing, versioning, and provenance. Applications of the technology include, but are not limited to: simultaneous computer aided design of molecules in 2D or 3D in which users may be distributed across multiple computers and in which the need for computer time synchronization (offline or online editing) is obviated; version control and provenance tracking of a virtual molecule; and other types of data used in computer aided molecular design activities.