Tuesday, July 23, 2019

Advances in AI and ML are reshaping healthcare

The healthcare technology sector has given rise to some of the most innovative startups in the world, which are poised to help people live longer, better lives. The innovations have primarily been driven by the advent of software and mobility, allowing the health sector to digitize many of the pen and paper-based operations and processes that currently slow down service delivery.



More recently, we’re seeing software become far more intelligent and independent. These new capabilities — studied under the banner of artificial intelligence and machine learning — are accelerating the pace of innovation in healthcare. Thus far, the applications of AI and ML in healthcare have enabled the industry to take on some of its biggest challenges in these areas:

  1. Personal genetics
  2. Drug discovery
  3. Disease identification and management

Upon close evaluation of the opportunities that exist within each area, it becomes obvious that the stakes are high. As such, those that are first to market with a sustainable product differentiation and value-add will benefit tremendously.

Ushering in a new era of personal genetics

The most significant application of AI and ML in genetics is understanding how DNA impacts life. Although the last several years saw the complete sequencing of the human genome and a mastery of the ability to read and edit it, we still don’t know what most of the genome is actually telling us. Genes are constantly acting out of place in combination with other variables such as food, environment and body types.

If we are to understand what influences life and biology, we must first understand the language that is DNA. This is where ML algorithms come in and the advent of systems such as Google’s Deep Mind and IBM’s Watson. Now, more than ever, it has become possible to digest immense amounts of data (e.g. patient records, clinical notes, diagnostic images, treatment plans) and perform pattern recognition in a short period of time — which otherwise would have taken a lifetime to complete.

Businesses such as Deep Genomics are making meaningful progress in this realm. The company is developing the capability to interpret DNA by creating a system that predicts the molecular effects of genetic variation. Their database is able to explain how hundreds of millions of genetic variations can impact a genetic code.

Once a better understanding of human DNA is established, there is an opportunity to go one step further and provide personalized insights to individuals based on their idiosyncratic biological dispositions. This trend is indicative of a new era of “personalized genetics,” whereby individuals are able to take full control of their health through access to unprecedented information about their own bodies.

The technology must have access to vast amounts of data in order to better curate lifestyle changes for individuals.

Consumer genetics companies such as 23andMe and Rthm represent a few of the first movers in this domain. They have developed consumerized genetic diagnostic tools to help individuals understand their genetic makeup. With Rthm,  users are able to go one step further and leverage the insights produced from their genetic test to implement changes to their everyday routine through a mobile application, all in real time.



As is the case with any application of AI/ML, the technology must have access to vast amounts of data in order to better curate lifestyle changes for individuals. Startups that are focused on mastering the delivery of personal genetics are doing so by considering the following key activities, as highlighted by Japan-based researcher Takashi Kido:


  1. Acquiring reliable personal genome data and genetic risk prediction
  2. Conducting behavior pattern analyses on people’s attitude to the personal genome to determine what kind of information is valuable/helpful and what type of information is damaging
  3. Data mining for scientific discovery

The second point is interesting in that not all genetic information about a patient’s biological predispositions is productive. Being able to control the information in a manner that is conducive to psychological well-being is critical.

Hyper targeted drugs are the future

Another exciting application of AI/ML in healthcare is the reduction of both cost and time in drug discovery. New drugs typically take 12 to 14 years to make it to market, with the average cost hovering around $2.6 billion. During the process of drug discovery, chemical compounds are tested against every possible combination of different cell type, genetic mutation and other conditions relating to a particular ailment.

As the task of doing this is time-consuming, this limits the number of experiments or diseases that scientists can look to attack. ML algorithms can allow computers to “learn” how to make predictions based on the data they have previously processed or choose (and in some cases, even conduct) what experiments need to be done. Similar types of algorithms also can be used to predict the side effects of specific chemical compounds on humans, speeding up approvals.

San Francisco-based startup Atomwise is looking to replace test tubes with supercomputers during the drug development process. The company uses ML and 3D neural networks that sift through a database of molecular structures to uncover therapies, helping to discover the effectiveness of new chemical compounds on diseases and identifying what existing medications can be repurposed to cure another ailment.

In 2015, the company applied its solution and uncovered two new drugs which may significantly reduce Ebola infectivity. The analysis was completed in one day — as opposed to years, which is common using traditional methods of drug development. A recent study by Insilico Medicine solidified the approach Atomwise is taking, showing that deep neural networks can be used to predict pharmacologic properties of drugs and drug repurposing.

The application of AI/ML in healthcare is reshaping the industry and making what was once impossible into a tangible reality.

Berg Health, a Boston-based biopharma company, attacks drug discovery from a different angle. Berg mines patient biological data using AI to determine why some people survive diseases, and then applies this insight to improve current therapies or create new ones.

BenevolentAI, a London-based startup, aims to expedite the drug discovery process by harnessing AI to look for patterns in scientific literature. Only a small portion of globally generated scientific information is actually used or usable by scientists, as new healthcare-related studies are published every 30 seconds. BenevolentAI enables analysis on vast amounts of data to provide experts with insights they need to dramatically expedite drug discovery and research. Recently, the company identified two potential chemical compounds that may work on Alzheimer’s, attracting the attention of pharmaceutical companies.



As advances in ML and AI continue, the future of drug discovery looks promising. A recent Google Research paper notes that using data from various sources can better determine which chemical compounds will serve as “effective drug treatments for a variety of diseases,” and how ML can save a lot of time by testing millions of compounds at scale.

Discovering and managing new diseases

Most diseases are far more than just a simple gene mutation. Despite the healthcare system generating copious amounts of (unstructured) data — which is progressively improving in quality — we have previously not had the necessary hardware and software in place to analyze it and produce meaningful insights.

Disease diagnosis is a complicated process that involves a variety of factors, from the texture of a patient’s skin to the amount of sugar that he or she consumes in a day. For the past 2,000 years, medicine has been governed by symptomatic detection, where a patient’s ailment is diagnosed based on the symptoms they are displaying (e.g. if you have a fever and stuffy nose, you most likely have the flu).

But often the arrival of detectable symptoms is too late, especially when dealing with diseases such as cancer and Alzheimer’s. With ML, the hope is that faint signatures of diseases can be discovered well in advance of detectable symptoms, increasing the probability of survival (sometimes by up to 90 percent) and/or treatment options.

The opportunities continue to grow and inspire healthcare practitioners to find new ways to enhance our health and well-being.

Freenome, a San Francisco-based startup, has created an Adaptive Genomics Engine that helps dynamically detect disease signatures in your blood. To make this possible, the company uses your freenome — the dynamic collection of genetic material floating in your blood that is constantly changing over time and provides a genomic thermometer of who you are as you grow, live and age.

When looking at disease diagnosis and treatment plans, companies such as Enlitic are focused on improving patient outcomes by coupling deep learning with medical data to distill actionable insights from billions of clinical cases. IBM’s Watson is working with Memorial Sloan Kettering in New York to digest reams of data on cancer patients and treatments used over decades to present and suggest treatment options to doctors in dealing with unique cancer cases.



In London, Google’s  Deep Mind is mining through medical records of Moorfields Eye Hospital to analyze digital scans of the eye to help doctors better understand and diagnose eye disease. In parallel, Deep Mind also has a project running to help with radiation therapy mapping for patients suffering from neck and head cancer, freeing up hours of planning for oncologists to allow them to focus on more patient care-oriented tasks.

What does all of this mean?

The application of AI/ML in healthcare is reshaping the industry and making what was once impossible into a tangible reality.

For AI/ML to become pervasive in healthcare, continued access to relevant data is essential to success. The more proprietary data a system can ingest, the “smarter” it will become. As a result, companies are going to great lengths to acquire data (which resides in an anonymized format). For example, IBM bought out healthcare analytics company Truven Health for $2.6 billion in February 2016 primarily to gain access to their repository of data and insights. In addition, they partnered with Medtronic to further Watson’s ability to make sense of diabetes through gaining access to real-time insulin data.

As the data becomes richer and the technology keeps advancing, the opportunities continue to grow and inspire healthcare practitioners to find new ways to enhance our health and well-being.
Share:

0 comments:

Post a Comment