How ChatGPT will transform medicine this year

Technologists have long discussed the limits of artificial intelligence (AI), arguing everyday on whether it will ever replace humans, a global discourse that is now cutting across all sectors, including medicine after US-based AI research company OpenAI launched ChatGPT-3, a query messaging service last November.

You give it a question and it answers in the most sufficient way it can, better than Google, and even offers five versions of the response.

I used it to search “how to diagnose blood cancer” and it listed five key ways — blood test, bone marrow aspiration and biopsy, lymph node biopsy, imaging test and genetic testing. It ends with a disclaimer that reads: “It's important to note that a diagnosis of blood cancer can only be confirmed through examination of a tissue sample. Your doctor may use one or more of the above methods to determine if a biopsy is necessary. Early diagnosis and treatment of blood cancer can improve outcome and survival.”

That is just one among billions of medical answers for many health questions stored inside this artificial brain. Its accuracy and contextualised responses mean health practitioners, by assessing the condition of a patient, can use it to deliver better diagnosis and treatment starting this year.

Though still in its initial phase, ChatGPT is already cutting down the time needed to conduct medical scientific research. I agree there are a number of things that need to be improved and further discovered by the platform, but it holds the biggest potential in transforming the global pharmaceutical industry and healthcare industry. Actually, when put to test by researchers last week, the generative AI appeared to pass the US medical licensing exam.

According to chief medical scientist at Microsoft Junaid Bajwa, medical knowledge doubles every 73 days, and with ChatGPT learning everyday about what has been published in medical journals and scientific papers, users of the platform can benefit from more specific, personalised, and result-backed healthcare solutions, treatments and consultations.

And with Kenya’s tech ecosystem booming, where over 130 health tech startups have sprouted over the last seven years, expect local innovators to use the tool to develop better telemedicine solutions, including medical apps that understand local vernacular languages. That means in future, your grandmother can get a surgical operation done on her by a robot that receives commands from, say Madrid, in Spanish but within milliseconds explains the procedure in Dholuo or Kikuyu so she can hear.

With ChatGPT, health administrators can now announce an official end to unnecessary scanning of paperwork because they can use it to pull out vital patient data from a centralised system, like finding a patient’s insurance information before a medical examination.

Hospitals can also use it to maintain a 24-hour customer care service, offering personalised responses to patients’ questions with accuracy and relevance. The technology also has a huge promise in improving hospital workflow and creating order in the rather disorderly emergency and ward rooms of countries like Kenya.

This means it can improve efficiency in the pharmaceutical supply chain by automating the process of generating purchase orders, invoices, and delivery notes which can save time and reduce the risk of errors like shipping the wrong medicine.

Many doctors are known to scribble medical notes that have too much jargon. This AI can translate the acronyms that commonly pervade clinical notes into words that are more commonly known to patients.

Its potential could also see the manufacture of new medical equipment this year and the coding of better medicine software to help in the treatment of diseases and health conditions that have been difficult to diagnose and treat such as celiac disease, cancer, fibromyalgia, HIV, multiple sclerosis, Alzheimer’s or Huntington’s disease. It can generate instructions for a new wearable device, which can help patients with chronic conditions to self-manage their conditions more effectively. This tech can generate protocols for clinical trials of new medical devices, which can help to speed up the development and approval of new treatments.

Doctors can use it in anesthesia to develop clinical decision support systems, deliver pre-operative education, and assist with the management of post-operative pain.

Using the voluminous medical data on ChatGPT, researchers can discover new drugs, providing a paradigm shift towards drugs that arrest diseases faster and with less side effects.

They are already leveraging deep learning and biomedical machine reading via natural language processing (NLP) models, with the target of achieving advanced self-supervised learning this year, like conducting task-agnostic biomedical language model pre-training and proposing a general framework for task-specific supervision. 

OpenAI has announced it will launch ChatGPT-4 by the end of this year which will be supported by 100 trillion parameters, so expect all glitches in ChatGPT-3 to be resolved as the technology moves to artificial general intelligence (AGI) where machines become more intelligent than humans.

The writer is the Africa tech correspondent at Quartz. [email protected]