Somewhere in our life we have, unfortunately, heard a doctor express with utmost despair that they could have done better if the patient could have been brought to medical attention a little early-on. Google thinks it can solve this problem for mankind.
Last year at Google I/O, Sundar Pichai demonstrated the advanced computing powers of Google named Tensor Processing Units (TPU) which now has been working with eye hospitals like Arvind and Shankara in India to help doctors use Deep Learning, a machine learning module to screen Diabetic Retinopathy in a better way.
Turns out, the AI they built can do more than just help in screening the retina, it can predictively detect if the patient is having chances of getting a heart attack or stroke; all this just by scanning the retina.
Such discoveries of AI’s capabilities have led Google to work with partners including UC San Francisco, Stanford Medicine, and The University of Chicago Medicine to develop a program that can help doctors predict medical events and help them answer tough questions like ‘is my patient likely to get better?’, ‘when can they be discharged from hospital’ and more importantly ‘will a patient get sick again’ and can they know it before the patient’s health condition deteriorates.
Data in a patient’s record is represented as a timeline for Google’s Deep Learning model to analyse it and predict chances of readmission.
Google has published a paper on how they are predicting medical event by using machine learning to analyse patients Electronic Health Records (EHRs). Google’s AI analyses about 100,000 data points per patient, something humans can’t do, to determine if their health is likely to deteriorate in near future and if they may need readmission. The current level of accuracy in prediction by Google’s AI is better than traditional methods by 10%; the aim is to empower doctors by nearly 48 to 24 hours in advance before a patient falls sick again.
The medical event prediction exercise was done by Google on a retrospective data, and the results are early; but a great start for machine learning and AI to help mankind live better and longer.
It is natural for us, lesser mortals, to wonder if such advanced AI can replace doctors? The answer is no and the reason is that, Google’s deep learning models are better listeners, their ability to predict chances of medical events stems from their capability to read and crunch large amount of data including diagnosis, medications, and notes created by master diagnosticians who we otherwise call as doctors!
Google aims to make healthcare better by helping doctors be more efficient and for patients to get better quality of healthcare, what I saw at Google I/O 2018 this is only a beginning of man and machine working together to push the boundaries of humanity.
Sreeraman Thiagarajan is co-founder of Agrahyah Technologies and listed at Google Developers Expert for Marketing. He tweets at @sreeraman.