CLINICAL DEVELOPMENT IN ONCOLOGY A
s such, generative AI (i.e. AI capable of creating text, images, videos, or other data using generative models)
is transforming nearly all aspects R&D. The McKinsey Global Institute (MGI) estimated that generative AI could contribute about $100 billion a year in economic value for the pharma and medical-product industries, largely because it can boost productivity by accelerating the process of identifying compounds for possible new drugs, speeding their development and approval, and improving the way they are marketed (Viswa et al, 2024). Companies not applying these tools may otherwise lose the competitive positioning among other players in this field (Bangul khan 2023; Viswa et al 2024).
ML is based on (deep) neural network (DNN)
interfaces, which mimic the nature of collecting and assessing data (similar to the connections between neurons and synapses in the human brain). This enables them to build decision trees, by probabilistic trial and error learning through input data and associated outcome modelled data. In oncology drug development, it could generate designated lead antibody compounds on an existing library of example compounds. This could give better certainty as to whether a company’s existing compounds are competitive, even before these compounds are being tested in
The McKinsey Global Institute (MGI) estimated that generative AI could contribute about $100 billion a year in economic value for the pharma and medical- product industries, largely because it can boost productivity by accelerating the process of identifying compounds for possible new drugs, speeding their development and approval, and improving the way they are marketed
various time and labor intensive invitro and animal models. Computational predictive toxicology tools
could aid in safe pharmacology testing of monoclonal antibody-derived therapeutics as they include elements of automation, consistency, and reliability to standard toxicological assays (Kizhedath 2017). Moreover, researchers can incorporate inhouse biomarker data with external data from collaborators, or open-source biomarker data to build a combined database to generate a suitable dataset which can be applied to a DNN learning model to inform on early efficacy response, well before it is tested in patients. Consequently, this will speed up the development process, significantly lower associated costs, and lower the number of patients required before entering pivotal trials. But there is still a huge gap in the
translatability between invitro data and animal models or humans (Seyhan 2019). For most common chronic diseases, it is known that they exhibit a complex and variable disease course, as well as different responses to therapy. In pharmaceutical development however, there is obviously a desire to capture the most relevant target being involved in any particular disease progression. To find a drug which is likely to have an impact on this target in humans requires finding the most causal upstream target or protein.
Difficulties in clinical trial organisation One of the pitfalls of the pharmaceutical industry in general, and in immune-oncology in particular, is that prediction of an immune- mediated response following immune therapy in different indications depends heavily on the individual status of the patients’ immune signature, underlying genotypes, mutations, pre-treatments, and of course the staging of the tumor itself. Applying potential DNN models using existing biomarker data from all patients in a company’s database and correlating with radiological response (such as iRECIST1), time of progression and other data would give better insight into future predictions of response. Importantly, different patient profiles, demographics, and underlying medication needs to be taken into consideration. In that context, access to
Clinical Trials in Oncology | 13
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44
