AI application in Healthcare

Continuing our series on AI implementation in different areas, today we’re talking about a field which has a huge impact on our lives — healthcare.

To the average person the level of change and sheer transformation that AI brings to the industry is not yet obvious. But we will see huge changes soon, in all aspects of healthcare from diagnosis and doctor interactions to medical therapy and health improvement.

This post can’t cover all possible AI and machine learning applications, let alone mention all the emerging technology and research that promises so much.

AI for better diagnostics

The advancements in image recognition and deep learning are making medical diagnostics more accurate.

Many talented teams are working to fight a variety of cancers through early diagnosis using AI. 2017 was a very successful year for achievements in this field; among them the ability to far more accurately detect skin and breast cancer by using AI to analyze medical scans.

Researchers from the John Radcliffe Hospital in Oxford, England, successfully developed technology using AI to analyze heartbeat images to detect heart attacks and other heart disease. Tests show that the Ultromics AI diagnostic system is more accurate than human doctors at diagnosing heart disease.

But accurate diagnose isn’t limited to image recognition. AI and machine learning are being used to perform odor analysis. The Owlstone company is a provider of chemical sensors and odor recognition technology that were successfully tested for diagnosing colorectal cancer and lung cancer.

The technology is based on the metabolite pattern recognition of breath samples (or samples of other biospecimens). The pattern, called metabolom, is a good indicator of human health. Changes in the metabolome signal a health condition, disease or inflammatory process. Correctly recognizing the pattern allows for early detection at a pre-cancerous stage where the survival rate is much higher.

Owlstone is now researching how this technology can be used for diagnosing other types of cancer and inflammatory diseases.

There several other companies working on similar technology. Hopefully, the competition will bring more products to the market faster, thus saving more lives.

AI implementation and trends in healthcare

AI application for drug discovery

Discovery of new drugs is a long and extremely expensive process. The challenge of drug discovery is that  from the thousands identified only some prove to have therapeutic value. Of those only a few are developed and only a small percentage of those make it through clinical trials to the market.

AI is changing all that. Advanced deep learning techniques are able to generate molecules structured with specified properties. The approach offers two valuable lines of research:

  • decreasing side effects among existing drugs
  • creating completely new drugs for a variety of difficult, even incurable, diseases.

There are huge advantages to using AI for drug discovery. For example, natural language processing can analyze massive amounts of patent data from different fields recognizing subtle relationships between molecules and compounds. By extrapolating that knowledges scientists see new links and develop new hypotheses. This is a very promising trend. Researchers tend to focus on specific diseases, often ignoring promising applications that don’t “fit.” AI encourages more cross-pollination between unconnected fields of research leading to unanticipated breakthroughs. 

AI application in genetic science

The industry, providers, medical diagnostic manufacturers, and drug companies, believe the future of medicine is a personalized approach, in which the diagnosis and treatment are based on patients’ individual genomic information.

The advancements in AI technology, along with the progress in computational power, are powerful tools for genome researchers and scientists.

In December 2017 Google announced the release of DeepVariant, an open source, deep learning technology to reconstruct the true genome sequence from high-throughput sequencing data, with significantly greater accuracy than previous  methods.  See details here.

Other fields of genetic science, such as genome editing, gene modifying and genetic engineering also benefit by utilizing AI. Better predictive models and more precise gene editing protocols improve the outcomes of experiments in fundamental research.

We know AI is an excellent tool for early diagnosis. Now AI can help map the individual genome, understand why certain mutations occur, as well as control the process of gene modifying. The combination opens the door to finding better treatments and possibly even cures for previously incurable diseases. And, hopefully, it will lower the cost to the point that these techniques will be widely available.

AI application in Population Health Management

Population Health Management (PHM) is another field where AI adoption yields multiple benefits.

AI-infused management platforms aim to optimize workflow in hospitals and automate back office operations, such as tax payment and parts of the supply chain. And as AI performs more of the medical staff’s routine daily tasks, such as scheduling, paperwork, data entry, etc., it will leave more time for direct patient interactions.

What is even more valuable for patients is that AI can improve medical insights. The deep analysis of data helps doctor to recognize best practice of medical care for certain conditions, predict possible side effects of medical therapy, and prevent some complications after surgery.

Traditionally, medical practice is based on case experience of what worked and what did not. Today, the sheer quantity of exponentially growing data makes it impossible for any human to stay current. Using AI for researching available treatments helps doctors make more informed decisions, as well as supporting the chosen treatment.

There are still a lot of tough question regarding AI implementation for population health management, such as privacy, cyber security and  acceptance of AI by physicians. These are the key factors slowing the progress AI can bring to the industry. According to a PWC report only 39% of healthcare provider executives say they are investing in AI, machine learning and predictive analytics.

AI in healthcare_pwc_report_2018

Hopefully, the just-announced partnership between Amazon, Berkshire Hathaway and JPMorgan, will showcase how adding tech in the right way can provide better patient outcomes at substantially lower costs, which could encourage/force faster adoption of AI in the healthcare arena. See details here

Final thoughts

Better diagnosis. Better treatment. Better customer experience.That is what AI and machine learning can bring to healthcare. No one expects to replace doctor or researchers, but with the help of AI everybody wins.

Healthcare has the reputation of being a tech-resistant field. Between the complexity of implementation and the regulations to ensure safety, total adoption could take as long as 20 years.

Hopefully, with the help of leading tech companies, this traditionally long process will be accelerated.

Leave a Reply

Your email address will not be published. Required fields are marked *