CLINICAL DEVELOPMENT IN ONCOLOGY


industry-wide databases would help generate better and more holistic knowledge (HMA-EMA Big Data Steering Group 2023). Within operations, the traditional way of


conducting a trial is bringing the patient to the hospital/trial unit. That is often cumbersome for patients, who must often travel long distances from home. This is especially common in early phase trials, when most procedures are being done at the site. For many clinical oncology trials, innovative approaches are being proposed, i.e. using remote monitoring, de-centralized imaging, and home nursing facilities near to patient homes. Modern technologies for real time assessment such as wearables or wireless devices for automated patient monitoring are already being used to ease logistics, make data transfer quicker, more accurate, and avoid data lags or incorrect data being collected. Patient predictive enrichment can reduce the number of patients needed for screening using digitalized histopathology images, single cellor spatial omics data enabling detection of mutations or transcriptomic profiles. In that sense, clinical research organizations (CROs) also need to apply new models to support such studies. This will require extensive digital data integration, data standardization across trial sites within and across countries, and harmonization of ethical and regulatory considerations.


Challenges of AI integration Whenever a pharma company or CRO wants to integrate AI and ML into its processes, it is crucial to first carefully assess what would be the appropriate methodology and research questions. One cannot simply apply the same ML algorithm from one platform to another because of differences in the data applied, differences in patient’s profiles, and disease etiology. Only when the methodology has been built on scientific well-defined grounds, one can ensure the reliability and validity of the model outcome. There are various metrics applied on the performance used in DNNs or supervised learning models between predicted and ground truths, albeit it is acknowledged that AI models outside of the settings in the lab in which they have been developed may differ in performance in a real word setting.


14 | Clinical Trials in Oncology


Only when the methodology has been built on scientific well-defined grounds can we ensure the reliability and validity of the model outcome


Additionally, performance evaluation and monitoring of AI models in clinical trials are currently limited. However, this is expected to quickly improve in future. Enabling diversity in patient data is crucial


to build reliable ML models on. For instance, a lack of racial diversity in a clinical trial data learning set can make incorrect predictions on entire populations. Data protection concerns must be


addressed and carefully built on each patient’s consent, otherwise AI in healthcare might be inaccurately used for commercial purposes. Non-integration and non-consensus at all


levels will not enhance acceptance of AI in clinical practice. European and US Health authorities paved the way for general ethical and standardization rules in this context. (EMA 2023, FDA 2023-2024).


Conclusion Even though pharma faces its own unique challenges, including complexity of biology, heterogeneity of patients and diseases, and an evolving regulatory environment, it is obvious that the future will be vastly affected by using AI in biomedical R&D. Patient responder analysis during clinical trials supported by AI and ML can leverage various data domains and potentially find early signs of treatment response, making trials faster, shorter, and less costly. The scientific and academic research community, industry, policymakers, and public in general must be open to the idea of integrating more inter- institutional, multidisciplinary collaborations. Otherwise, emerging economies including China, India, and Korea with their unique R&D and business models will outperform the long-held leadership of US and European pharmaceutical companies.


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