Artificial intelligence is being increasingly used across different industries, from identifying legal defences to transferring money for banking customers. But what is it? Defined by Dictionary.com as:
“The capacity of a computer to perform operations analogous to learning and decision making in humans, as by an expert system, a program for CAD or CAM, or a program for the perception and recognition of shapes in computer vision systems.”
This Stanford University paper defines AI as:
“The science and engineering of making intelligent machines, especially intelligent computer programs. It is related to the similar task of using computers to understand human intelligence, but AI does not have to confine itself to methods that are biologically observable. Intelligence is the computational part of the ability to achieve goals in the world. Varying kinds and degrees of intelligence occur in people, many animals and some machines.”
Big data in the life sciences and healthcare sectors is an increasing trend. It is the processing and displaying of huge volumes and varieties of data at a rapid speed. There is a vast amount of data in the healthcare and pharma industries, including lab data, insurance data, patient records, research data, and even social media data. But true AI is when a system can understand and learn from complex sets of data whilst making valid recommendations and suggestions.
Pharmaceutical companies have vast amounts of compounds that could be the perfect solution to combat specific diseases, but they have no way to identify them as such. The development and production of drugs can cost pharmaceutical companies up to $2.6 billion (£1.8bn) and take 12 to 14 years to complete. Thus, the main short/medium-term implication AI has for the pharmaceutical industry is the reduced time it takes to develop drugs and thus the associated costs, enhancing return on investment and could even mean a reduction in cost for end users.
A Boston based pharma start up, BERG uses a data driven approach to drug discovery. This is the company where maths meets biology, and the firm believe this combination is key to discovering the correct drugs for complex diseases.
Their software uses AI to process huge amounts of biological data to reveal what happens (in minute detail) in the journey from good health to cancer. This allows better insights, more informed hypothesis, and thus more efficient drug development.
The AI program processes a huge amount of data. Aside from the genomic information, one single cell tissue could produce over 14 trillion data points.
Using this approach, the team discovered the key role mitochondria had in allowing cancer cells to flourish in pancreatic cancer. Cancer cells can turn off the mitochondria which results in them losing their ability to die. BERG’s drug BPM 31510 helps the mitochondria function and essentially turns the cancer cells back into normal cells. The team believes they can use this model for other forms of cancer.
This bioinformatics firm has taught its AI system to predict the therapeutic use of new drugs before they even enter the testing process. The AI processes huge amounts of data from experiments on human cells using known drugs. Over time, the AI program was able to achieve roughly 55% accuracy identifying one out of 12 of the drugs therapeutic applications. Although this seems like a small number, it is actually a big step for researchers who would have to do many experiments (both time consuming and costly) to reach the same conclusions.
Through machine learning and hypothesis testing, The University of Manchester’s ‘robotic scientist’ Eve recognises why it has succeeded and thus improves at the tasks it needs to perform. Eve is designed to automate early stage drug development and has already discovered lead compounds against malaria and African sleeping sickness.
Houston Methodist Research Institute
Researchers at Houston Methodist Research Institute have developed an AI model that can predict breast cancer risk, allowing doctors to closely monitor those at future risk. Currently, around half of mammograms yield false results, according to the American Cancer Society, and means evasive procedures are sometimes unnecessary.
The program interprets mammograms and translates patient data into diagnostic information 30 times faster than a doctor, and with 99% accuracy. Manual review of 50 charts took two clinicians 50 to 70 hours, whereas the AI software reviewed 500 charts in a few hours, saving the human doctors 500 hours of their time. It scans patient charts, collects diagnostic features and correlates mammogram findings with breast cancer subtypes. With this information, clinicians used the results to accurately predict the patient’s probability of breast cancer diagnosis.
Processing large clinical and medical data isn’t the only place where AI could affect the pharma industry. Even business and marketing based decisions could be helped by computing ‘brains’, for example by analysing and assisting with mergers and acquisitions and providing guidance on the most efficient and effect way to market new products.
Although AI is in its infancy, the fusion of maths, technology, biology and healthcare is becoming more and more talked about in the industry and progress in being made in developing systems with real-world value. There is wider talk of AI being an existential threat to humanity (Stephen Hawking, Bill Gates and Elon Musk have all voiced their words of warning), but in the specific application of pharmaceuticals and treating diseases, then any increase in efficiency and effectiveness can only be welcomed.