AI and Healthcare

No comments

By Taneshq Verma

Artificial intelligence in Healthcare, simplified as use algorithms that can mimic human cognition in analysis of complex medical reports and data. With increase in computation power and amount and complexity of medical data of a person have forced us to intervene the predictive power of algorithms in healthcare also. As we are planning to replace car drivers by autonomous vehicles, same way some laborious works of a doctor majorly – a radiologist, can be replaced by some predictive algorithms by some extend.

The role of AI in healthcare can be impacted by AI as – one using AI to make more powerful instruments, medical devices and second in prediction, diagnosis or prognosis of a disease. Third getting to know about the genetics of the disease and then formulate the meds accordingly. As we go in the history of use of AI in healthcare, it goes back to 1960s, project Dendral was started in late 1960s in Sandford, whose aim was to study hypothesis formation and discovery in science. Organic chemistry was the topic chosen initially, where they tried to identify molecules by analysis of the compound’s mass spectra and knowledge of chemistry.

The use of AI is enabling review and translation of mammograms 30 times faster with 99% accuracy, reducing the need for unnecessary biopsies[1]. IBM Watson Health is exploiting the concepts and practices to use AI for healthcare. Watson for Drug Discovery is leveraging AI to help researchers uncover new insights for combating Parkinson’s Disease. As we know the predictive algorithms are generally data hungry, the big names experimenting with AI in Healthcare field have taken care of that in their own ways. Like the Watson Platform for Health accesses 200 million claims, 100 million de-identified patient records plus other genomic and social determinants of health data. Google’s DeepMind Health is also working in partnership with clinicians, researchers and patients to solve real-world healthcare problems. The technology combines machine learning and systems neuroscience to build powerful general-purpose learning algorithms into neural networks that mimic the human brain, this is slightly deviated from our topic but in the long run it will be the solution to most of the healthcare problem. The world’s most complex machine, our brain, the most complex and yet to be understood. If someday we are able to study the brains of the creatures then we can also alter them. That day we can also train the brain to cure diseases itself, take the other regenerative power from lizard gene and train that in a human brain or alter the human gene accordingly to have a regenerative power.

AI will help us to study the brain and Google’s DeepMind and Elon’s Neuralink are just stepping stones for this kind of future, where we just have to download a software of how to cure a disease of be genetically immune to a lot of deadly diseases.

Using AI doctors can have a head start on life-threatening diseases and they can also have a predictive prognosis of an ill person and prepare accordingly.

Also most of R&D are happening for diagnosis of the disease. Google’s DeepMind has is also focusing on prediction of acute kidney injury as there are about 1M cases alone in the UK every year. These all early diagnostics will help us to move from reactive to preventive healthcare in upcoming future. 

In this article I would be stressing on the later part which includes diagnosis, prediction and prognosis of disease with the help of computation power in the present world and machine learning algorithms. After reviewing the evidence from the last few decades, Journal of American Medical Association has marked the incidence of cancer misdiagnosis between 10% and 20%, misdiagnosis outweighs both drug errors and botched surgeries as an avoidable cause of patient harm.

And in a country like India where we are facing a shortage of 6,000,000 doctors as per UN norms, we can exploit the diagnosis of diseases with the help of computers and not much trained doctors and reduce the misdiagnosis cases by a huge number.

Companies like PAIGE.AI are doing a lot in diagnosis of cancer with the help of AI, Computational pathology and clinical experts. They are working on general and organ-specific modules. PAIGE will be able to fulfill tasks ranging from rapid diagnostic stratification to tumor detection, segmentation, prediction of treatment response and survival. They have lots of publications in this field that are available on their website. They are working on algorithms that takes petabytes of data. Their introductory video gives a very clear view of the big picture which Thomas wants to convey. Companies like this works on hard core machine learning algorithms. And many of them still being used as a black box.

Then we have startups like Willtok – they are data-driven, enterprise SaaS company that delivers the healthcare industry’s leading consumer activation platform – as per their website. They generate data at large scale of 270M and 800 parameters and then apply analytics and predictive algorithms to identify individual needs, receptivity and impact ability. They also have collaborated with IBM Watson for applied artificial intelligence in healthcare.

In the near future with this increasing R&D in this field we can see ourselves been treated by computer end to end with minimal human interference. Starting from diagnosis going till surgery all things can be automated.

These all are industry level solutions of problems related to healthcare. There are a lot of competition in which people from educational institute solve the healthcare issues leveraging artificial intelligence. The research groups are very niche in this field but is growing day by day.

Talking about the impact on mankind, the misdiagnosis of diseases the poor prediction of prognosis, mostly due to human error or the specific cases that can slip through a human  being can be taken care of by the computational power, and predictive algorithms that can learn from petabytes of data and millions of parameters, with high confidence interval.

Taneshq Verma is a tech analyst at Vistas News

Leave a Reply