Unlocking Gene Therapy with DataHowLab
Overcoming Cost and Accessibility Challenges
Gene Therapy holds the promise to revolutionize medicine by managing the root cause of genetic diseases and unlocking the cure for otherwise untreatable conditions. To deliver on that promise, many challenges remain to reduce the development costs and make the therapies more affordable to patients.
A complex manufacturing process and high costs pose a challenge for development
A significant challenge in Gene Therapy processes lies in their complex, sequential nature, where each step is highly interdependent and must be carefully controlled to ensure the final product meets stringent quality requirements.
Unlike traditional biologics, where a single phase must be controlled before final critical quality attributes (CQAs) are measured and assessed, the sequential nature of gene therapy processes result in significantly higher complexity and room for error, increasing the risk of failed batches, costly rework, and extended development times.
Rework costs further escalate the already high costs associated with gene therapy development. The extensive resource demands (material, cost, and time) in turn limit the availability of experimental process data. Scientists are therefore challenged to develop highly complex processes while trying to minimise experimental effort.
DataHowLab enabling a consistent, effective digital development approach
DataHow’s AI-driven technologies are proving to be instrumental in overcoming these barriers by delivering crucial development insights while optimizing processes with a digital-first approach. By leveraging hybrid modeling and transfer learning techniques, DataHow is enabling a more efficient, data-driven, and scalable path to gene therapy development.
Underlying DataHowLab Technologies: Hybrid Modeling & Transfer Learning
Hybrid Modeling
Hybrid modeling is ideal for the unique constraints of bioprocessing, particularly in gene therapy development, where data availability, process complexity, and high process variability pose significant challenges.
DataHow’s hybrid models seamlessly integrate mechanistic modeling—which algorithmically represents first-principles process knowledge—with data-driven machine learning. The mechanistic foundation captures known process dynamics, reducing the scope of uncertainties. Machine learning then focuses only on the remaining unknowns, enabling more accurate predictions even with limited experimental data. This complementary approach enhances process understanding, even when faced with limited data
Transfer Learning
Transfer learning is an advanced machine learning methodology that harnesses historical process data to transfer knowledge across to new developments. This approach enables faster process understanding and optimization while remaining resource light. As insights are transfered from historical data, experimental burdens reduced, associated costs are minimised, and development time are shortened – all critical parameters for gene therapy.
Assessing the Impact of DataHowLab on Gene Therapy
To illustrate the impact that DataHowLab and its technologies has on gene therapy development, a selection of industrial cases are highlighted:
Case 1: Assesing the ability of DataHowLab to accurately understand and predict final CQAs when provided a limited number experiments from a gene therapy development.
Case 2: Assessing DataHowLabs ability to estimate the impact of key process parameters accross each phase of a gene therapy process to enable the development of a more targeted control strategy.
Case 3: Assessing whether data from historical projects can be used to improve process understanding for novel gene therapy developments.
Key Results
DataHowLab and its advanced technologies are transforming gene therapy development. Despite the inherent complexity of gene therapy processes, DataHowLab delivers deeper, more holistic process insights, enabling a more efficient and data-driven development pathway. For the full results with supporting plots, download the full case study below:
Case 1: DataHowLab’s hybrid modeling approach demonstrated significant analytical efficiency, accurately understanding and predicting final CQAs with only 12 experiments. Furthermore, the models where able to accurately predict across all phases of the process – growth, perfusion, and transfection.
Case 2: DataHowLab empowers scientists with actionable insights into the relative impact of key process parameters on each critical quality attribute (CQA) at every phase of production. By quantifying variable importance at each stage, the platform provides relevant intelligence that enables targeted, phase-specific corrective measures, accelerating convergence toward optimal process conditions, and the development of targeted control strategies.
Case 3: By integrating datasets from two historical gene therapy process with the target project, DataHowLab significantly enhanced process understanding and reduced prediction error for the target process. This data-driven approach reducies the reliance on costly and time-consuming experiments—a critical factor for gene therapy development.
Should you wish to further explore the case with the DataHow team, please contact us directly.
