Yesterday, Bill Gates released his End of Year report, speculating how artificial intelligence (AI) will evolve while also reflecting on becoming a grandfather for the first time.
An annual event, the Microsoft founder spent time focusing on how AI is being used to change the world of medicine and healthcare.
“Artificial intelligence is about to accelerate the rate of new discoveries at a pace we’ve never seen before,” Gates wrote on the GatesNotes blog.
“One of the biggest impacts so far is on creating new medicines. Drug discovery requires combing through massive amounts of data, and AI tools can speed up that process significantly.”
In addition to highlighting how the Gates Foundation is looking at how these tools can address health issues like AIDs, TB, and Malaria, Gates highlighted some promising use cases, including using AI to combat antibiotic resistance, treating high-risk pregnancies, assessing people’s risk for HIV, and documenting medical records.
Using AI to Combat Antibiotic Resistance
Among the healthcare use cases highlighted by Gates was the work of Nana Kofi Quakyi, a program manager at the Aurum Institute in Ghana, who is developing an AI model to help healthcare workers prescribe antibiotics without triggering antimicrobial resistance.
Antimicrobial resistance, where pathogens develop the ability to resist antibiotics if used too much, is a significant public health concern, which the World Health Organization (WHO) claims was “directly responsible” for 1.27 million global deaths in 2019.
In response, Kofi Quakyi is working on an “AI-powered clinical decision support tool that allows prescribers to enter prompts, respond to system queries, and receive personalized, real-time antibiotic prescribing recommendations, such as drug, dosage, and duration of administration.”
The model is trained on a dataset comprising clinical guidelines, research data, expert opinions, and curated health surveillance data. The goal is to identify which pathogens are at risk of developing resistance and suggest treatment options to the patient based on this information.
Treating High-Risk pregnancies
Another ambitious project highlighted in the post involves researchers at the Indian non-profit ARMMAN, who are developing a large language model (LLMs) to act as a virtual copilot for healthcare professionals treating high-risk pregnancies.
High-risk pregnancies are a significant global issue, with the United Nations (UN) reporting that a woman dies during pregnancy or childbirth every two minutes.
ARMAN’s copilot, available in English and Telugu, is designed to provide greater contextual depth so that medical practitioners can better understand how to treat these patients and reduce mortality among expecting mothers.
Automatically Assessing the Risk of HIV
HIV is a global health issue that affects 39 million people around the world. However, as Gates mentioned, Sophie Pasco of Wits Health Consortium (Pty) and others in South Africa are looking to use an app to help patients assess their risk of HIV.
The app, known as Your Choice, enables users to interact with an LLM-driven chatbot, which collects details about their sexual history and uses it to improve HIV risk assessment.
Your Choice has the potential to provide 24/7 support for patients to use if they are unable to discuss their sexual history with their doctor and is intended to extend support to “marginalized and vulnerable populations.”
Making Medical History More Accessible
Legacy approaches to storing patients’ medical records often make it difficult for healthcare professionals to get up to speed on a patient’s medical history.
This makes it challenging to identify if an individual has any allergies or underlying healthcare conditions, which could affect treatment going forward.
That being said, Gates notes that Maryam Mustafa, Assistant Professor at the Department of Computer Science at the Lahore University of Management Sciences in Pakistan, has been developing a voice-driven LLM that can fill out a patient’s medical record.
Essentially, the app uses technologies like generative AI, speech recognition, and text recognition to ask questions about the patient and then uses the responses to produce a standard medical record.
Gates’ End of Year report highlights that AI, particularly LLMs, plays a transformative role in the healthcare industry. The examples above show how language models can be used to improve patient care and medical professionals’ access to insights, which can help manage treatment options.
Over the next year, we can expect more similar pilots to undergo development as medical professionals look to automation to help save and improve lives.