Healthcare AI Boosts Drug Discovery, Medical Training, Wearables And Concerns

Artificial intelligence (AI) has been propelling every industry it touches, and the healthcare industry is no exception—nor should it be.

Especially considering that AI technology helps boost productivity, improve turnaround time for research and development, reduce costs, and cut down time taken by routine and menial tasks.

It can also help in smoothening and quickening the process of drug discovery, which is exactly what Washington-based data-first AI drug creation company Absci did.  

Absci’s GenAI-Designed Drug

Recently, the company announced that the first healthy volunteers have been dosed in a randomized, double-blind, placebo-controlled, first-in-human Phase 1 study meant to evaluate the effectiveness of the company’s ABS-101 drug—for the Treatment of Inflammatory Bowel Disease (IBD)—one that’s been engineered from scratch using Absci’s generative AI (GenAI) platform.

Through the study, Absci’s team will evaluate safety, tolerability, pharmacokinetics (PK), and pharmacodynamics (PD) in around 40 healthy adult volunteers.

"We are excited to advance ABS-101 into Phase 1 clinical trials,” said Andreas Busch, PhD, Absci Chief Innovation Officer. “Designed and optimized using our generative AI platform, ABS-101 exhibits high affinity and potency, expected low immunogenicity, and an extended dosing interval—all of which are intentional attributes achieved through AI."

As per the company, GenAI played an instrumental role in the development of the drug, helping the company reduce its development time from five years to just over two years—or half. It also brought down costs from the typical $50 to $100 million to just over $15 million.

Combining Absci’s 77,000+ square foot wet lab with its GenAI prowess allows the company “to predict antibodies from scratch that can bind to a target of interest,” as per Sean McClain, Founder and CEO of Absci.

“With a growing pipeline of AI-designed biologics, we’re accelerating a new chapter to bring better biologics to patients, faster,” McClain added.

Mount Sinai’s GenAI Training

The use of GenAI in healthcare is a serious avenue for the revolution of the industry.

So much so that the Icahn School of Medicine at Mount Sinai in New York City has begun incorporating AI into its doctor training program, becoming the first school in the nation to do so.

As part of the move, Icahn is granting access to OpenAI's ChatGPT Edu to all its M.D. and graduate students. According to Faris Gulamali, a student using the tool, ChatGPT helps him prepare for surgeries and better explain complex diagnoses to patients.

Even OpenAI—the creator of ChatGPT—is taking it extremely seriously.  

OpenAI’s GenAI Benchmark - HealthBench

“Improving human health will be one of the defining impacts of AGI,” said OpenAI in a blog post introducing HealthBench, a new benchmark designed to better measure the capabilities of AI systems for health.

“If developed and deployed effectively, large language models have the potential to expand access to health information, support clinicians in delivering high-quality care, and help people advocate for their health and that of their communities.”

The new benchmark has been built in partnership with 262 physicians who have practiced in 60 countries, speak 49 languages, and have training in 26 medical specialties. It also comes with 5,000 realistic health conversations, each with a custom physician-created rubric to grade model responses.

The tool exudes OpenAI’s belief that AI healthcare systems should be meaningful, trustworthy, and unsaturated, i.e., support progress and enable continuous improvement.

OpenAI also published the performance of its models based on the HealthBench benchmark across seven themes—emergency referrals, responding under uncertainty, health data tasks, global health, expertise-tailored communication, context seeking, and response depth—all with evaluation examples, candidate responses, and rubric criteria & grades.

The AI leader also said it uses HealthBench to evaluate how recent frontier models perform and chart progress over the last few years—the company found that recent OpenAI models have improved rapidly across frontier performance, cost, and reliability.


Of course, GenAI will improve wearable devices.  

Fitbit’s GenAI Medical Records Decoder

While rumors of Fitbit’s upcoming demise fly around, reports find that Google is looking to empower it with Gemini’s AI capabilities.

Along with a new round of Labs that users can preview, Fitbit is bringing Medical Records Navigator to users soon, which will help them “turn complicated lab reports into clear health summaries.”

The app—currently in waitlist—will also allow users to “securely upload recent lab results to the Fitbit app” and use Gemini to extract data, summarize information, and present the data to users in clear, accessible, and easily understandable language, complete with educational resources.

Fitbit is also previewing two new Labs—Symptom Checker to let users describe symptoms and get possible reasons after completing follow-up questions, and Unusual Trends to “track changes to your health data or overall well-being.”

Despite all these advancements, GenAI can’t seem to get rid of basic but important concerns. 

Foresight’s GenAI Training Concerns

While the makers of the GenAI model Foresight call it a boon in predicting diseases or hospitalization rates, others seem to question the methodology by which it has been empowered to do so. 

Reportedly, the tool has been trained on the medical data of 57 million people who previously used the services of England’s National Health Service.

In 2023, England’s population was estimated to be almost 57.7 million.

The case has caused a massive stir in the country since even the developers of the tool can’t promise the tool won’t disclose sensitive patient data through its responses.

The current version of Foresight, powered by Meta’s Llama 2, uses eight different datasets of information collected by the NHS from November 2018 to December 2023, spanning outpatient appointments, hospital visits, vaccination data, and records.

The initial version of the tool, which was powered by OpenAI’s GPT-3, used data comprising 1.5 million patient records from two London hospitals.

What do you think about the recent AI advancements made in the healthcare sector? Do you think GenAI’s capabilities justify and outweigh its concerns?

Let us know in the comments below!