Artificial Intelligence To Treat Cancer?
By Sandhya Mishra
Sandhya Mishra talks to the experts on much buzzed artificial intelligence swooping in Healthcare, executing diagnosis to treating cancer
This is the era of AI (artificial intelligence) revolution that has penetrated into every bailiwick, healthcare not being an exception. Healthcare in India on the other hand is handicap with acute shortage of medical practitioners, infrastructures thus subsuming the innovation more than ever. Max healthcare, Apollo and other hospitals and many startups are already making news for employing AI in tidbits.
Now what’s breaking the internet are the speculations that AI could possibly treat a cancer. With Microsoft buying data from SRL Diagnostic to train its AI to diagnose Cancer, it is predicted that days are not far when robots will be treating patients in future.
Artificial intelligence is claimed to ‘learn’ from the massive database of patient’s medical data to make fool-proof diagnosis, interpret the prognosis, make treatment protocols and what not
Anil Bhansali, Managing Director, Microsoft R&D Pvt Limited
Growing prevalence of chronic diseases such as cancer is one of the key factors for the need for pathology to transform. Digital pathology is increasingly being preferred as it has helped bring improvement in service delivery, patient safety and communications while reducing errors and lowering costs. Histopathologists require highly efficient tools to assist in diagnosis, thus augmenting the demand for automated and innovative implementation of cloud and AI.
Microsoft’s collaboration with SRL Diagnostics will help improve the quality of digital pathology for population screening by bringing together Microsoft’s Azure and AI innovations along with SRL’s world-class expertise in the study of cells and tissues (histology). With this addition, Microsoft’s AI Network for healthcare has now expanded to include pathology along with eyecare and cardiology, thus representing the continued efforts made by the organization to democratize healthcare.
AI Network for healthcare thus empowers the healthcare providers with faster, intuitive and predictable solutions and reducing the disease burden. We are passionate to make further advancements in the healthcare sector. We are looking at creating a common data language and architecture that would make healthcare services easier and simpler to engage with, improve the speed and predictability of care, and support the delivery of precision medicine that ultimately saves more lives.
Arindam Haldar, CEO, SRL Diagnostics
The AI platform that Microsoft will build is expected to create a software environment infused with millions of data points and knowledge gained from SRL’s expert laboratory professionals.
Anatomic Pathology, the current mainstay in cancer diagnosis, is highly human expertise dependent and cannot be instituted in a way that routine pathology testing works. Since the whole test review process is manual in nature, starting from scanning the slide (or an image of it) to identifying a suspicious area and making a diagnosis, the process is inherently time consuming. Furthermore, the final diagnosis is often dependent on the experience and the expertise of the reporting pathologist.
AI application can make Digital Pathology far more intuitive by adding a layer of intelligence in biopsy slide imaging. Initial slide scanning, identification of tissue type and areas of interest can be done by an AI solution. The AI system learns from the tens of thousands of data points already in the system under pathologists’ supervision. When such an AI system applies this machine learning into new slides, it reduces the room for error while also significantly improving efficiency and speed. However, human acumen and intuitiveness will be vital in making the final pathology diagnosis in a case, but at a faster clip as the preliminary steps would be automated A faster pathology result would logically mean a faster institution of therapy.
Artificial intelligence and machine learning have the potential to become the greatest asset in improving healthcare services especially in India, where there is just one doctor for around 1,700 people in India and about 70 per cent of healthcare infrastructure is focused around few cities that cater to only 30 per cent of the population. According to Frost and Sullivan’s report, the AI market for healthcare applications is expected to achieve rapid adoption globally, with a CAGR of 42% until 2021. While a lot is happening on AI front globally, in India there is a white canvas, presenting a huge opportunity and a sea of possibilities.
Artificial Neural Networks and Natural Language Processing capabilities of super computers may revolutionise Clinical Diagnostics in coming years and will enable Big Data analytics based services. Artificial Intelligence can use algorithms to ‘learn’ patterns from a large volume of healthcare data, and then use the obtained insights to assist reporting of diagnostic tests for use in clinical practice. With the growing number of samples of cancer, as well as of other diseases, there’s a need to quickly and accurately analyse the samples to help doctors arrive at a diagnosis faster. These capabilities of AI are surely going to impact healthcare in a big way.
Dr Sachin, Consultant Medical Oncologists, P.D. Hinduja National Hospital
Some pathologists like all other human creatures are likely to make human errors when diagnosing a problem while many others are very good at their job and this difference is mostly because of the volume of cases latter has seen.Such kind of errors may get reduced by using technological advancement like an Artificial Intelligence that can validate your findings. However, it’s too early to talk about the future of AI in healthcare
One thing is for sure that no matter how much data a machine is made to ingest it can never become better or even equal to human doctors. How can machine break the diagnosis to a patient? Treating patients is definitely more than diagnosing, setting treatment protocols and medicating them. AI led diagnosis can be used to verify the diagnosis once made but alone can’t be held fool-proof, even the most touted robotic surgeries have limitations. There are also issues regarding sharing of patient’s data to Developers or other institutions considering privacy of patients getting snapped.