ChatGPT and Healthcare
Alex Nubla
Executive Leader / CIO / CTO / Innovator / Technology / Strategy / Cloud Transformation / Digital / AI / ML / Data Science / Program ? Product Delivery / BCP ?DRP
Many have played with #ChatGPT by now. Some asked ChatGPT to write poems or even a song for DRAKE (I actually asked it to write me a song like Taylor Swift.)?Before I provide you the 411 or go deeper dive in ChatGPT, one must understand how it responds based on the data it has been trained (or collected/stored). ChatGPT is very useful for research, pulling the information for you based on the collection of data it has stored. ChatGPT rely heavily on NLU (Natural Language Understanding) and NLP (Natural Language Processing) model trained on humongous dataset collected for decades. It means it can understand text and reply through text similar to how humans do. Like your voice assistant Siri. 10 years ago during it's inception, when I ask Siri "Call me a Taxi", Siri would reply "Gladly Alex, starting today, I will call you TAXI!" Fast forward to today and NLU has changed.
So how will we utilize #ChatGPT in the Healthcare space. If you ask #ChatGPT "How will healthcare change in the Future?" - the replies will be based on the collective data of healthcare trends that authors have written in the past years.?It will talk about AI, Machine Learning, Telemedicine, IOT that will revolutionize the healthcare industry.?Other topics will include Data-Driven Healthcare information within EHR/EMR or expanding your health information through Blockchain technology, more Patient Centric Care - listening to the Voice of the Patient, Value based healthcare where providers are more focus on the quality of care instead of quantity of services, growth of Telemedine for adding multi-channel care available to patients, and even growing AI based analysis on patterns of your visits - like determining whether you'll be diabetic based on your OCT scan and how it can be reduced / prevented based on the medication and care provided. It is important to note that these are existing journals written up in the healthcare industry and all ChatGPT did was collect all these data.
Reducing the research hurdles for clinicians and IT appear to be the place where ChatGPT will make an impact. I used to have to depend on Gartner and Forrester research to unveil products in the market that will align with our IT strategy.?Here's an example, 5 years ago when researching for NLU for the IVR modernization that can execute Conversational AI - we have to research all through Google. Not to mention each vendor may have different terms for what they meant Conversational AI to be. Today, I just asked #ChatGPT "What is the best NLU product in the IVR market today?" The most common products for major vendors popped up - Microsoft LUIS, Nuance, IBM Watson, Google Dialogflow and Amazon LEX. These are advertized market where the Vendors have the means and funding for marketing the product. It's unfortunate that smaller market vendors may not get a good voice of their product. Similar to what Google does today, if you don't pay for their service - you SEA / SEO will likely not be discovered.
领英推荐
For Clinicians, where sometimes it is imperative to find answers immediately. #ChatGPT is a superb tool to help support and provide assistance to a diagnosis. However, this is just supplemental documentation to provide physician productivity. Don't loose touch of what you learn and what you have done in years past. #ChatGPT, like in IT, is a resource to provide physician with information.??
For Patients, #ChatGPT may serve as a breakthrough for getting information - but like I tell my Uncle, it does not replace a Doctor's advice or consult (my Uncle thinks GOOGLE is his doctor and GOOGLE knows it all - I have not introduced him to ChatGPT yet.) The key to this innovation is how healthcare leaders utilize this asset.?#ChatGPT is a way to improve patient lives, but it will not replace what our healthcare staff can do. #ChatGPT lacks emotions and empathy. Just like NLU / NLP - it processes Text based conversation. Reads through what dialog is entered, searched through the Data Lake and replies based on what is available in Storage.?Your reply will depend on the Data stored - 1. Is the data Complete? 2. Is the data Accurate? 3. Is the data available Now??Overall, #ChatGPT will still depend on the data stored and collected. Like AI categorization - there are "predefined" and "anamolies" categorization.?Part of AI is determining the anamolies and moving them to the predefined state - that's what makes your data collection rich.??
#ChatGPT is a work in progress. It will not replace humans. The usage in healthcare must have considerations on Privacy, Security, Accuracy, and informed consent from a patient. We can place good governance on the data itself, but too much restrictive regulations will also change the outcome of how #ChatGPT replies. It will be interesting in the upcoming years how our industry utilizes #ChatGPT - I know it made my research less stressful, but the bigger market vendors have a large stake on how replies are done in #ChatGPT. You must still be diligent in your research.
I think it will have a difference on scheduling and pre-op/pre-doctor care as people will be able to get more answers taken care of or addressed before even stepping into a facility.