Sorry, you need to enable JavaScript to visit this website.

The Rise of Personalized Medicine with Predictive Analytics

The Rise of Personalized Medicine with Predictive Analytics
April 30, 2019

Recently, one of my closest friends shared some surprising news about her daughter Tara. The seven-year old had just been diagnosed with Type 1 diabetes. For a child, this means a life without ice-cream, cakes or candies, and for her parents, the endless worry and vigilance that goes with such a chronic disease.

Unlike Type 2 diabetes, which can be managed with weight management, Type 1 diabetes is a chronic autoimmune condition. With Type 1 diabetes the immune system attacks the insulin-producing beta cells in the pancreas which triggers diabetes. The exact causes are unknown and there is no known cure. The only way to survive this condition requires the patient to overhaul their lifestyle, undergo frequent blood sugar tests, regularly take insulin and other medications, do daily exercises, and plan their meals carefully. It’s a life changing condition.

Thankfully for Tara, modern technology had an answer that made their lives a lot easier. A small sensor that continuously monitors Tara’s blood sugar levels and provides early notifications on oncoming fluctuations throughout the day. More than once, this sensor has helped Tara and her parents address sudden changes ahead of time and avoid a trip to the hospital because the young child ate some ice-cream.

Such sensors didn’t exist even 10 years ago. Today, these sensors allow doctors and healthcare providers to analyze the medical data patterns and make better decisions on how to better manage day to day healthcare. Such advancements demonstrate the power and potential of data driven personalized medicine. And there are many more advancements just around the corner.

Biggest Challenge to Predictive Personalized Healthcare

Imagine a world where sensors, like Tara’s, can do more than just monitor glucose levels. Armed with real-time sensor data, a full medical history, genetic and lifestyle information, the possibilities are endless. Every individual empowered by a personal healthcare system that can not only predict diseases but also give real-time healthcare guidance to improve their overall quality of life. Such technology can save lives of millions, if not billions across the world.

But such a technological revolution requires vast quantities of data from across multiple platforms and healthcare systems to be possible. We have already taken the first steps on this journey over the last few years with the digitization of healthcare data using electronic health and medical records (EHR/EMR). According to the Healthcare Information and Management Systems Society (HIMSS), the volume of this data is increasing at an exponential rate of 48% a year. As any data scientist will tell you – more data means more challenges.

It’s no surprise then that the Society of Actuaries reports that 16% of providers believe that the main obstacle to implementing predictive analytics in healthcare is “too much data”. Moreover, this problem is deeply rooted in the healthcare system as payers, providers, researchers, patients, and vendors, all maintain highly siloed data sets. In my experience, ensuring the interoperability of data across disparate silos remains the single largest challenge faced by most next-gen healthcare providers and enterprises. However, there is progress being made as organizations begin to adopt a value-based model and realize the benefits of data interoperability.

Personalized Health through Predictive Analytics

It is clear to me that advances in machine learning (ML), artificial intelligence (AI), and massive parallel genetic sequencing will lead the way to personalized healthcare. After all, just look at how we use these technologies in other parts of our lives. Digital video platforms like Netflix use AI and ML to analyze diverse audience behavior and preferences to offer personalized content and improve customer experience.

By combining real-time data about audience preferences and viewing habits, Netflix is able to offer recommendations, suggestions, and alerts for content. This personalization extends to the very first user layer of the platform where no-two people are likely to see the exact same customer menu. Healthcare is no different, just a lot more complicated.

But modern technologies are working steadily to solve these complicated challenges. Ever since the mapping of the human genome, our understanding of personal health has skyrocketed. We have applied this new knowledge to healthcare systems in a multitude of ways – from identifying the relationship between ailments and genes to assessing the impact of lifestyle factors on genetics.

In 2017, the University of Pennsylvania developed a predictive machine learning tool using EHR data to identify patients most likely to suffer from sepsis or septic shock almost 12 hours in advance. Similarly, analytics based clinical-decision support tools developed by CancerLinQ and the FDA have successfully used genetic information and previous patient studies to predict patient response to various therapies, allowing them to make optimal choices.

Combined with AI and ML, our ability to apply medical knowledge across vast population data sets can now generate valuable new insights. These meaningful insights will have a direct impact on patient health outcomes with better visibility into care management, lower costs of care, better supply chain efficiencies, patient experience, patient safety, drug efficacy and many more such benefits.

Combined with AI and ML, our ability to apply medical knowledge across vast population data sets can now generate valuable new insights.

Predictive Healthcare Can Transform the Industry

On an industry wide scale, predictive analytics has the potential to enhance value-based care while also driving costs down. While descriptive analytics has largely been used so far, the shift to predictive analytics promises incredible returns for both payers and patients. For example, a 2017 report by the Society of Actuaries states that about 60% of healthcare executives believe predictive analytics will save them 15% or more in costs over the next five years. And this is just the beginning.

Another report from the Society of Actuaries tells us that almost 75% of all healthcare costs are rooted in 17% of high cost patients. These high cost patients mainly suffer from one or more chronic conditions which can have a remarkable toll on national healthcare spend. In fact, a 2018 report from the Milken Institute says that the total healthcare costs for chronic healthcare conditions amounted to 5.8% of the GDP at USD 1.1 trillion in 2015-16. Additionally, if we account for other indirect costs such as lost income and slowdown in economic productivity, the total economic cost of chronic disease was nearly $3.7 trillion or nearly 20% of the GDP during the same period.

The Health Care Cost Institute (HCCI) has already been successful in using predictive analytics to identify future high cost patients. Payers now need to rethink their business models and use these insights to focus on lowering healthcare costs in the face of changing customer behavior. In fact, as per another Society of Actuaries report, nearly 92% of payers and providers agree that predictive analytics is essential to the future of their business. Since predictive analytics will allow them to identify high cost patients, they need to devise various data driven solutions that can incentivize them to lead healthier lifestyles.

The importance of addressing the health needs of potentially high cost patients cannot be understated. Predictive analytics can make personal health a gamified experience that uses health data along with behavioral information to improve lifestyle and act as a preventative measure. Gamification has already shown some promise in other areas and can be deployed by insurance and healthcare providers to help patients engage proactively with their own treatment. This is particularly important since healthcare is not only about how conditions are managed within hospitals but also about an individual’s own accountability towards their health outside of hospitals.

A Better Future with Analytics

As value-based care becomes the norm, next gen enterprises need to leverage predictive analytics to improve patient experience, optimize population health, and reduce the per capita cost of healthcare. If pursued relentlessly, predictive analysis in healthcare will make medical treatment as convenient as ordering groceries online or hailing a cab from an app. I imagine a shift in the healthcare paradigm that will make medical treatments function more like the current retail experience. With greater insights into patient behavior and needs, healthcare will operate just like Amazon, Walmart or Uber does today.

The convergence of technologies like IoT, cloud, analytics and even 3D printing, will result in a world where transplant organs go from being a previous rarity to a bespoke product, manufactured for an individual’s need on-demand. I wouldn’t be surprised if in the next 20 or 30 years, we witness an Uberization of healthcare when all complex in-patient treatments will be managed in the comfort of people’s own homes via some advanced form of telemedicine.

Perhaps this transformation will take longer than 20 or 30 years, but it is a reality that is well within our grasp. At least, until then, we can aspire for the next step in that direction and bring in a new healthcare paradigm where children like Tara do not have to fear hospitalization for just having ice cream. And predictive analytics in personalized healthcare is the pathway that will lead us to just such a future.

Reference:

  1. https://www.himss.org/event/big-data-big-management
  2. https://www.soa.org/Files/programs/predictive-analytics/2019-health-care-trend.pdf
  3. https://www.newswise.com/articles/machine-learning-may-help-in-early-identification-of-severe-sepsis
  4. https://healthitanalytics.com/news/cancerlinq-partners-with-fda-for-precision-medicine-studies
  5. https://www.soa.org/Files/programs/predictive-analytics/2018-health-care-trend.pdf
  6. https://www.soa.org/Files/resources/research-report/2018/2018-predict-high-cost-hcci.pdf