Those using computer science to study biology have had little issue raising capital in recent years to bring their ideas directly to market. The promise was clear: if you can get your hands on big datasets, machine learning will be able to overcome persistent issues with drug discovery, development, and testing.
However, a lot of these pioneering drug programs have faltered. It is critical that patients and the field comprehend the reasons behind these developing aches and find solutions. To date, we’ve identified five major obstacles to success and provided solutions.
- Patient data is essential
Using cutting-edge expertise from an academic lab to address a technical challenge with well-defined datasets is a tried-and-true strategy for success in the tech and biotech industries. However, the use of artificial intelligence (AI) in medicine, sometimes referred to as “techbio,” is really more about “bio” in its broadest definition than it is about “tech.”
Within the complexity of numerous healthcare settings, the data needed for success is typically more varied, coming from more sources, and—most importantly—originating from real people.
- Remaining near the academic realm
Human intelligence must come both before and after artificial intelligence in the creation of medicine. It is found in research hospitals and universities. Thus, establishing strong, long-lasting alliances with academic institutions and research hospitals is essential to maximizing the potential of artificial intelligence to improve healthcare.
- Multimodal data are necessary for multidimensional situations
We can state with confidence that, at this point in the medical field’s use of AI, AI has succeeded in providing us with a fresh perspective on human biology.
For instance, a single digital pathology slide has as much data on it as a full-length movie, and machine learning is the only way to fully utilize it all. However, it is crucial to accept the complexity of biology at all scales, from molecules to cells, tissues, the disease microenvironment, and the organism as a whole, in order to comprehend disease and be able to alter its trajectory.
- Boost the likelihood of clinical trials
In clinical trials, the great majority of experimental treatments are unsuccessful. This is extremely costly for both industry and society. Patients receive fewer potentially effective therapies, and the expense of these failures depletes the funds that could be used to launch new initiatives in the numerous unmet need areas.
The way the trials are set up is one of the main causes of this. In a perfect world, medications would only be tried on patients who had specific ailments and would address the underlying biology of those conditions. Then, clinical trials might be extremely modest and almost always beneficial. In actuality, our knowledge of illnesses, drug interactions, and the precise disease type of trial participants is far from optimal.
More so than we could have predicted even a few years ago, AI can assist us in completing these stages. We are able to map entire disease processes for the first time thanks to machine learning. That will eventually lead to innovative new treatments.
- Regulators as collaborators in new ideas
In addition to its enormous promise, artificial intelligence (AI) presents regulatory problems related to the possibility of bias in algorithms, the proper use of personal data, and, most crucially, guaranteeing patient benefit.
Regulators who apply their cautious, risk-based methods to precision medicine innovation, such as the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA), are actively involved with these emerging technologies. It makes sense that a greater emphasis on AI/ML-powered medicine would lead to fewer clinical trial participants, higher approval rates, and reduced overall expenses.
