Assisted AI- New paradigm in healthcare
There was a significant development with respect to Artificial Intelligence and healthcare a couple of months back. FDA, for the first time, approved a software powered by artificial intelligence that could significantly aid a doctor to interpret medical imagery. The software is called IDx-DR, made by diagnostic AI startup IDx. The software analyzes images of the retina to detect a complication of diabetic retinopathy.
Diabetic retinopathy (DR) is non-symptomatic making it difficult to diagnose early
Why would this be a significant development? For starters here are some statistics. There are over 400M people who have diabetes, 145M have some form of diabetic retinopathy and 45M have vision-threatening DR. In India 15M people are blind but 80% of these could have been avoided, meaning they could have been prevented from going blind had there been an intervention early on. But this disease DR is non-symptomatic: One will never know until it is too late, unless there is periodic screening. On top of it, there are only 20,000 ophthalmologists in India.
15M people are blind in India, 80% of the cases could have been avoided
Replacing doctors is not an option. But I think artificial intelligence can aid doctors to make better decisions faster. It can also improve access as the doctor and the patient need not be in the same location anymore. This is in a way democratizing healthcare.
Forus health Co-founded by KC has partnered with the Andhra Pradesh (AP) government to run probably the world’s largest tele-ophthalmology program. This is connecting 13 districts and 115 community health centers of AP under a single roof using technology to do mass screening for the population.
Another application on AI that we are very interested is in the area of diagnostics. India’s diagnostic market is estimated to be $7B growing at about 17%. There are approximately 200,000 labs and collection centers in the country. And guess what? We only have 19,000 pathologists. Another interesting observation during clinical trials was that the intra-sample variability across practicing professionals can go up as high as 33%. This is a clear opportunity where AI can assist professionals to bring better access and quality.
There are 200,000 labs and collection centers in India; we have only 19,000 pathologists.
Sigtuple with their new AI 100 platform is trying to do exactly that. By employing image recognition and machine learning techniques, they intend to help clinicians look at samples of body fluids in a different light.
The earliest known examples of compound microscopes, which combine an objective lens near the specimen with an eyepiece to view a real image, appeared in Europe around 1620. Today with strong image processing software it is possible to digitize body fluid samples to great detail. With inferences from over 220M cell images and over 150B data points, the AI engine developed by Sigtuple is able to deliver Specificity greater than 99% and sensitivity above 98% for all classes of WBC identification. WBC DC correlated was within the bounds of inter-machine variability published in the literature.
Replacing doctors is not an option; Artificial intelligence can aid doctors to make better decisions faster.
The disease burden on the world is very complex and it is only increasing. The growth in the number of medical professionals, unfortunately, is not able to keep pace with the increase in new forms of medical conditions. The new forms of disease are more lifestyle related and hence less symptomatic making early stage diagnosis even more complex. The narrative that technology is going to replace the doctor is very far from the truth. We should look at technologies like AI as a tool to help clinicians improve productivity, lower the costs and deliver a better patient experience.
AI can positively impact early stage diagnosis of certain medical conditions, which could lead interventions early hopefully at lower costs leading to better outcomes for patients
As technology becomes more pervasive, there is bound to be more digitization. This allows for better audit trail on the diagnosis which will help in better understating of the effectiveness of treatment protocols and patient outcomes. Of course, there are big concerns on data privacy and countries all over the world are working on strengthening the laws around ownership of medical data and their transferability. But I am sure we will find answers to these too. It is clear that assisted AI can positively impact early stage diagnosis of certain medical conditions, which could lead interventions early hopefully at lower costs leading to better outcomes for patients.
India offers a unique combination of large data sets, good technology & clinical talent. India also challenges you to dig deeper into technology to meet the required price points while adhering to global compliance standards. We are excited to see an increasing number of folks trying to built solutions that address health issues that are globally relevant from here in India.
If you think you are doing something interesting please do reach me at ranjith_menon@idgvcindia.com
