This is the second in our blog series on “Digitizing the Human Body.” (Also see Part 1 and Part 3.) 

Artificial intelligence (AI) and advanced machine learning promise to reinvent healthcare, a sea change heralded by numerous AI-driven innovations, pilot projects, products and techniques. In fact, the use of AI in the life sciences sector is projected to explode, from $600 million in 2014 to over $6.5 billion by 2021. While Mark Zuckerberg, Elon Musk and others argue about AI’s ethical aspects, the innovations continue apace.

Here’s a quick view of some of the cutting-edge AI activities we’re involved with in the life sciences space.

  • Skin cancer detection: Cognizant has developed an AI-powered system that can analyze skin images and detect irregularities. By comparing a patient’s imaging results with a series of pre-stored images, the system helps to quickly and cost-effectively identify skin disorders. In our pilot trial, which covered five different disease tracks, the rate of detection accuracy was a little over 73% – thereby demonstrating great potential as a diagnostic aid.
  • Neurological disorders: Fans of the new CBS series Bull have seen how reading micro expressions can provide insights into understanding human behavior and be used to trigger specific reactions. But while legal dramas make for fine entertainment, an even more fascinating prospect is the neurological application of micro expression detection to identify early signs of health issues. We’ve built a system that can detect tiny variations in facial expressions, gait patterns and tremors by analyzing videos of individuals and detecting regions of interest in single video frames.  The system compares the key areas underlying human facial expressions (e.g., eyebrow and chin movements, lips and nose position) against a large store of pre-fed images that correspond with a wide range of human expressions. From this, the system makes logical deductions about an individual’s neurological condition. While not yet an industrialized solution, the system has the potential to go far beyond human powers of discernment.
  • Diabetic retinopathy: DR is one of the major causes of blindness today, with nearly 422 million people at risk. A diabetic patient’s retina looks very different from a normal person’s, as there are anomalies in existing blood vessels and sometimes abnormal growth in new blood vessels. We’ve built a system that applies a combination of neural networks and machine learning to enable medical specialists to compare pictures of the back of the eye – DR retinas against healthy ones – and thus determine the severity of the disease. The accuracy of this automated DR screening method is continuing to improve. Although our application has delivered strong results during its pilot stage, the system does not yet predict disease as accurately as an ophthalmologist can. But its current level of accuracy is high enough for the system to assist ophthalmologists in diagnosis and then match solutions to the disease state.
  • Oral cancer prescreening: Determining a patient’s predilection for oral cancer generally requires consideration of a vast array of predictive variables – lifestyle, family history, etc. Even then, full certainty is not possible, absent indicative data such as molecular imagery and pattern recognition. This is exactly where machine learning comes in. Our company is in the early stages of applying our knowledge of image analytics coupled with risk factor scores to assess risk levels for contracting oral cancer.

In Healthcare, AI Extends Human Touch

With all these advancements, there is a common misconception that embracing AI means removing the human touch from healthcare. To the contrary, AI’s uncanny ability to suggest the presence of human consciousness (Amazon’s suggestions are often right on the money, aren’t they?) makes it a welcome adjunct to the humans providing medical treatment. Far from replacing healthcare personnel, AI will imbue the human/technology healthcare continuum with a responsive quality. Ironically, AI and machine learning will ultimately offer the warm and compassionate care that patients seek – and boost health outcomes.

Further, AI will play a vital role in reducing medical errors and misdiagnosis. Andrew Beck, Associate Professor of Pathology, Harvard Medical School, has shown that AI can reduce misdiagnosis by up to 85% in the case of cancer detection – a huge margin, considering the number of cancer cases today. In the absence of a specialist, AI can also empower general physicians by assisting them in recommending the right treatment and in supervising patients. By doing so, AI can enable healthcare professionals to focus more on what matters most: providing the best treatment.

Contrary Factors

Of course, AI also introduces some level of risk and challenge that can’t be overlooked. As a conservative and regulated sector, healthcare will require more time to adapt to AI- and ML-assisted care. Another big challenge is collecting the right data that will lead to the best decisions. As Dr. Robert Mittendorff of Norwest Venture Partners says, “Curated data sets that are robust and have both the breadth and depth for training in a particular application are essential but frequently hard to access due to privacy concerns, record identification concerns and HIPAA.”

The benefits of AI in healthcare, however, far outweigh the risks and challenges involved. In a world that’s in dire need of lower cost, higher-quality and more accessible and personalized healthcare, it will soon be considered medical malpractice not to consider AI.

This is the second in a series of blogs on the burgeoning field of human digitization. Part 1 looked at emerging technologies that are ushering in an era of human augmentation, self-regulation, automated diagnosis and even vast intelligence gains. Part 3 looks at three FDA-related developments suggest regulatory hurdles are lowering when it comes to human digitization technologies in life sciences.

Pratik Maroo

Pratik Maroo

Pratik Maroo is Chief Digital Officer in the Life Sciences business unit at Cognizant. He leads Cognizant thinking in defining digital for... Read more