AI and life sciences: Navigating risks and challenges


Even as artificial intelligence (AI) continues to transform other sectors, the life sciences industry has only recently begun to realize its full potential. The reality is that AI-driven developments are quickly changing the face of the industry:

  • AI played a significant role in the development of viable COVID-19 vaccines in less than a year, a process that traditionally has taken 10 years or more.
  • AI recently discovered a novel drug candidate for liver cancer in just 30 days 1.
  • Google Cloud, which recently launched new tools powered by AI that promise to help the industry speed drug discovery, is just one technology company planning to launch technologies designed to automate previously time-consuming, manual processes2.


“We are at an inflection point,” said Dr. Almira Chabi, chief medical officer and chief development officer at HanAll Biopharma, who also serves on the Grants Review Group for the California Institute for Regenerative Medicine. “While there have been challenges and delays, the life sciences sector is adopting AI at remarkable speed and would expect to see significant acceleration of progress.”


In 2012, an article in Nature Reviews Drug Discovery3 discussed “Eroom’s Law” — Moore’s Law in reverse — in reference to the steep decline in pharmaceutical R&D efficiency. The authors noted that the cost of R&D for new drugs approved by the USFDA had risen exponentially for 60 years. By 2020, the same journal noted the start of change in the trend line within the prior decade. One of the factors attributed to the turnaround was better information for decisions, specifically better knowledge of disease biology or how to modulate it, leading to new ways to treat diseases and moving into diseases that lack effective therapies. Dr. Chabi notes that AI is already contributing to significantly increasing the availability of such information.


“As AI continues to evolve within life sciences, ethical considerations, data privacy, and regulatory challenges need to be carefully addressed to ensure responsible and beneficial implementation of these technologies,” said Grant Thornton Life Sciences Global Leader David Dominguez.



Futurist Jim Carroll on AI in life sciences



05:50 | Transcript


AI’s application range


AI has the potential to evolve nearly every stage of drug development, ranging from discovery to diagnostics, biomarker development to clinical trial design,” said Dr. Chabi. “With every incremental improvement, we can potentially decrease timelines and increase success rate, bringing therapies to more patients in need.”


Drug discovery and development is one area in which AI is having the greatest impact. Historically, the timeline from drug discovery to market launch has been an expensive, time-consuming and risky one, with the time from the discovery of a new drug to market typically taking as long as 15 years, with capitalized costs exceeding $2 billion and a success rate of less than 10%4.  Now, AI enables researchers to mine vast storehouses of digital patient data to speed the development of drugs and increase their chances of success.


In addition to drug discovery, AI is impacting other critical use cases:

  • Clinical trial design — The clinical trial phase of drug development, which often includes extensive and repetitive manual data collection and integration of data from disconnected sources, can take up to seven years. AI-powered algorithms can facilitate patient selection for clinical trials. AI-enabled data collection can significantly reduce the time and work needed to complete clinical trials. And AI-enabled study design can accelerate clinical trial protocol design.
  • In her role as chief medical officer for HanAll, Dr. Chabi is collaborating with the Michael J. Fox Foundation, NurrOn Pharmaceuticals and Vincere Biosciences, to explore leveraging data collected by the Parkinson’s Progression Markers Initiative (PPMI), a landmark study that has produced vast amounts of patient data over the years. The study has played a significant role in the design of new and emerging clinical trials. “PPMI data has been used, for instance, to develop algorithms that help stratify patients and predict which patients may be more likely to respond to a new drug,” Dr. Chabi said. This may make it possible to create a clinical trial design with greater precision and with a higher probability of success.”
  • Disease diagnosis and prognosis – AI can analyze large amounts of medical data and generate insights that can be used to identify diseases and develop personalized treatment plans. AI and machine learning have also proved to be adept at analyzing images and detecting features that would otherwise not be detected by humans. This has huge potential for use in the early detection of prostate and lung cancer,5 heart disease, diabetic retinopathy, and other diseases.
  • AI is quickly becoming an important tool in the early diagnosis of ophthalmic diseases, an advance that is important for ophthalmology, which is experiencing a shortage of trained professionals in many countries. However, Dr. Chabi anticipates that AI can provide a role beyond screening tools for detecting early-stage disease.6 She previously led a project leveraging deep learning for creation of algorithms correlating structure and function in glaucoma. Such tools have the potential to accelerate development of new treatments in this neurodegenerative condition7.


Headshot of  David Dominguez

“As AI continues to evolve within life sciences, ethical considerations, data privacy, and regulatory challenges need to be carefully addressed to ensure responsible and beneficial implementation of these technologies.”

David Dominguez

Life Sciences Global Leader


Industry series stories












AI’s risk landscape


While there is great potential for AI to transform the life sciences industry, there are many risks and challenges that remain.



Patient confidentiality


The use of patient health information is critical to the success of every innovation. Yet, because it relies on data collected from individual patients, ensuring the security of that sensitive data must be a high priority. That requires a focus on protecting the confidentiality of the patient data and ensuring that storage and transmission of that data is secure.


When companies work with many large datasets, anonymity can often be lost, which has been cited by many as a major concern. “Because use of large patient datasets is an ongoing trend in the industry, it is essential that data governance, data security safeguards, and a compliance program are in place as patient data is collected, stored, and processed—in order to ensure data security, privacy, and confidentiality,” said Dominguez. “By implementing key measures, organizations can significantly reduce the risk of data breaches and protect patient data effectively.”



Data quality and bias


The life sciences industry relies on vast quantities of data from patients, providers and payers — and ensuring the reliability of these datasets for generating insights can be a challenge. “AI models are trained on real-world data, which can contain biases,” Dr. Chabi said. “Once the biases are embedded in the algorithms and they are deployed, there can be significant consequences to patients.”


To ensure real business value from insights generated from AI models, it is important for life sciences companies to implement an effective dataset design. In addition to making sure they stay current on this fast-moving field, CEOs and top management at life sciences companies must encourage the implementation of processes that can mitigate bias, using best practices published by Google AI and other resources.8



Regulatory compliance


Regulatory requirements regarding patient safety and ethical practices are stricter than ever for the life sciences industry and the implementation of AI technologies raises the stakes. “When organizations do not meet these requirements, they face increased risks of legal consequences, financial penalties, reputation damage, and other negative impacts,” Dominguez said.


“An organization can minimize this risk by staying informed about relevant regulations, implementing a robust compliance program, conducting regular audits, establishing proactive measures, and fostering a culture of compliance and strong commitment within the company.”



Skills gaps and workforce challenges


One of the biggest barriers to advancing AI initiatives is the lack of skilled talent. According to a recent AI Skills Gap study, 93% of organizations in the U.S. and the U.K. consider AI to be a priority with planned projects, yet more than half (51%) say they do not have the right mix of skilled AI talent to progress their AI initiatives.


“Companies have been mitigating the shortage by forging partnerships and increasing collaborations across both industry and academia to share expertise and data.” Dr. Chabi said. “Biopharma R&D is transforming and greater flexibility, agility and cooperation are needed.”


More stories on artificial intelligence








The road ahead


Greg Corrado, co-founder of Google Brain, has described two phases of the AI revolution: the first phase was about teaching machines to recognize patterns. The second phase is the generative AI revolution currently occurring and it goes beyond pattern recognition to actual pattern completion9. According to Dr. Chabi, the life sciences industry is poised to greatly benefit from this second phase of the AI revolution.                                                                                                                                          







How AI is implemented in other key industries




Our life sciences featured industry insights