When will the real use case of AI come? It’s a million dollar question. People actually want to know if AI has attained so much intelligence, why it’s still far from offering any concrete solution in the field of medical science. While OpenAI is still focused on ChatGPT, Google is attempting a medical chatbot, something people actually want.
Google is currently conducting tests on an advanced artificial intelligence program that specializes in answering medical questions.
According to a recent report by The Wall Street Journal (WSJ), Google has been testing an AI tool called Med-PaLM 2, a medical chatbot that safely answers medical questions. It’s testing the product at renowned healthcare institutions like the Mayo Clinic research hospital.
Med-PaLM 2 is designed to harness the power of Google’s LLMs, aligned to the medical domain to more accurately and safely answer medical questions and is based on Google’s language model called PaLM 2.
According to the Google’s blog published in April, Med-PaLM 2 was the first LLM to perform at an “expert” test-taker level performance on the MedQA dataset of US Medical Licensing Examination (USMLE)-style questions, reaching 85%+ accuracy, and it was the first AI system to reach a passing score on the MedMCQA dataset comprising Indian AIIMS and NEET medical examination questions, scoring 72.3%.
Is passing the Medical exam enough?
Passing an exam does not necessarily make you a good doctor. It is just a criteria for the society to know that you have basic knowledge of your field and patients can trust you. In 2022, 91% of the candidates cleared the step 1 of USMLE exam according to the official data. Does it make all of them good doctors?
The expertise of doctors comes from the real time scenarios depending on patient to patient. Patients differ in their unique characteristics, and the process of prescribing medication cannot be generalized. Each patient’s body functions differently, and individual factors such as allergies must be taken into account. Google, being aware of this complexity, acknowledges the importance of personalized medical care.
The research paper Towards Expert-LevelMedical Question Answering with Large Language Models published by Google and DeepMind accepts the limitation of Med PaLM 2. The paper said “We note that our results cannot be considered generalizable to every medical question-answering setting and audience.”
The Med PaLM 2 is trained on multiple-choice and long-form medical question-answering datasets from MultiMedQA excluding patient’s personal data following ethical norms.
However, having a patient’s personal data will improve its efficiency to a whole new level but it is very much likely that patients will not be comfortable in sharing their health information as it is personal. Furthermore, Google executives confirmed customers testing Med-PaLM 2 would retain control of their data in encrypted settings inaccessible to the tech company, and the program wouldn’t ingest any of that data.
Should we ignore Healthcare LLMs?
Despite having limitations, use cases of Health care LLMs cannot be ignored. They just need to be handled properly as it is a matter of life and death.
This is no hidden fact that the medical field makes a lot of advancements with each passing day and it is tough for the doctors to be at the top of it.
Apart from Google, Microsoft is also making strides towards Healthcare LLM. The tech giant, which is also the biggest investor in OpenAI, in April, teamed up with the health software company Epic to build tools that can automatically draft messages to patients using the algorithms behind ChatGPT.
“Medical knowledge doubles every 73 days,” said Junaid Bajwa, chief medical scientist at Microsoft, in an exclusive interaction with Analytics India Magazine, at Global AI Summit, Riyadh. He said it is truly about the richness of information – the data coming from publications on medical research, particularly related to ailments and treatments for various diseases and medical conditions across the globe. Looking at the publishing rate of research papers, he estimates that it has the potential to double every three days in a few years.
This is where medical language models (LLMs) become valuable. Imagine a scenario where a doctor misses a critical approach during a medical emergency because they were unaware of an alternative treatment method. In such situations, LLMs can provide significant assistance as they possess a wealth of textbook knowledge. By accessing this extensive knowledge base, LLMs can help doctors by offering alternative approaches and ensuring that vital information is readily available when needed most.
Healthcare LLMs can assist doctors in having informative discussions, answering complex medical questions, and finding important information in difficult medical texts. With the doctor’s expertise and LLMs inputs , together they can make a formidable team.
Google told employees in April that an AI model trusted as a medical assistant could “be of tremendous value in countries that have more limited access to doctors,” according to an internal email reviewed by The Wall Street Journal that quotes a researcher working on the project.