The pharmaceutical industry has an enormous wealth of data from preclinical and clinical studies, genome sequencing and other sources. Artificial intelligence (AI) is able to sift through, analyze and process this wealth of data in the shortest possible time. This opens up completely new possibilities for research and development.
To make the most of these opportunities, Novartis founded an AI Innovation Lab with Microsoft in 2019. With more than 100 applications, AI plays an important role in all areas of our business. Here is an overview of where we use it:
Drug development
For Novartis, AI in research and development is an indispensable tool for data collection and analysis. It helps to predict the properties of molecules, to analyze protein structures and to identify patterns that would remain undetected by human data collection. Molecular simulations allow us to test the effect of compounds before they are transferred to clinical trials. This allows us to identify potential active ingredients more quickly. This is relevant insofar as the development of a new drug currently takes more than ten years on average and more than 95 percent of all drug candidates fail in clinical trials. AI can eliminate these doomed candidates at an early stage and focus on those preparations that have the greatest chance of success.
The ambitious MELLODDY project, in which Novartis was also involved, has shown how the use of data through AI can be taken to a new level for drug research: Several pharmaceutical companies have made their data available to train AI with machine learning, thus maximizing the potential of data for drug development while maintaining data sovereignty.
Cost minimization
The use of AI harbors huge savings potential. At Novartis, we therefore also use AI to optimize and automate processes and working methods, thereby minimizing costs. If we succeed, we can reinvest what we save by using AI in the research and development of innovative medicines and therapies.
Clinical studies
AI leads to the acceleration and improvement of clinical trials. On the one hand, it can help to identify suitable trial participants by comprehensively analyzing patient data, thereby shortening the selection process. On the other hand, it can support data analysis and the detection of patterns and anomalies, thus facilitating the implementation of clinical studies.
Diagnoses
Thanks to comprehensive data analysis, AI enables more accurate, faster and more accessible diagnoses. Novartis and Microsoft have used this advantage to develop an AI-powered system that can recognize leprosy based on photos of pathological skin changes.This allows people in the most remote places to photograph their skin, upload the image to the cloud and have it diagnosed by dermatologists anywhere in the world with the help of AI.
Treatments
When it comes to therapies, AI acts as a door opener to personalized medicine. The analysis of genetic information makes it possible to identify individual disease risks, reactions to medication and the optimal dosage in advance and to tailor therapies to patients. In addition, AI tools such as the We-Chat-Mini-App AI Nurse for heart failure patients developed by Novartis and Tencent promote patient engagement and empowerment by helping patients make informed decisions with data-based recommendations.
Ethical Risks
The potential of AI for the development of pharmaceutical data is countered by technical and ethical risks, such as poor quality, data protection violations or the bias of algorithms. A framework is needed to counter these risks. The Artificial Intelligence Act is currently being negotiated at EU level, but a fully applicable directive is not expected before 2026.
Until then, pharmaceutical companies must set a good example. Novartis has developed ethical guidelines to ensure the safe, transparent and responsible use of AI. Our measures include informing patients which personal data we use for which purpose in AI applications or using comprehensive and representative data for the development of AI algorithms in order to rule out bias.
About the Author:
Bettina Resl is a graduate of the Data Science in Health certificate program and is now a lecturer on the program. Here she teaches on the topic of "Public Affairs - Health Makes Policy: What does health policy have to do with health data?".