AI in Healthcare — The Real Opportunities

AI is hot again. Like clockwork, we emerged from an AI winter with a flurry of new technologies such as deep learning, cloud computing, and massively parallel GPU processing. Researchers make breakthroughs. Entrepreneurs launch new companies. Newspapers tell daily stories of amazing achievements. We truly are in a new golden age of Artificial Intelligence and Machine Learning.

Sure enough, healthcare is a major focus of these efforts. At $3.5 trillion (with a T!) and 18% of the U.S. economy, healthcare is ripe for innovation. Our aging population demands new methods of care. Our exponentially growing costs require efficiency gains and cost savings. Indeed, healthcare is a massive opportunity for AI.

But let’s take a closer look. What is real and what is hype?

Recent months have told a new story, one of skepticism. The lofty goals of curing cancer and robots replacing doctors seem as distant as ever. Ambitious researchers and entrepreneurs are confronted with the reality of distrust in black boxes, a complex regulatory environment, and messy unstructured data.

At One Medical, we have a unique viewpoint by virtue of physicians, engineers, and data all under one roof. We have the opportunity to realize the predicted benefits of integrated care. Our physicians and patients are not fooled by hype — we provide amazing care by solving real problems. Our collaboration with clinicians focuses our work on applications that measurably improve the healthcare experience. That vantage provides focus and clarity on where the real opportunities lie.

Here are the 4 pillars of One Medical’s AI and ML strategy. If you are an engineer, entrepreneur, or vendor seeking opportunities, these are for you:

Patient Experience

Our patients always sit at the center of our decision making. How can we provide the right care, at the right time, from the right provider? AI and ML dramatically enhance our ability to route patients. One Medical is famous for our same day appointments and high facetime with physicians. Behind the scenes, that requires herculean efforts to understand, prioritize, connect, and route.

Opportunities:

  • NLP to understand patient intent
  • Automated planning and schedule optimization
  • Recommendation engines for optimal care pathway
  • Patient engagement to manage chronic disease and preventative care

Provider Experience

Our providers — physicians, nurses, phlebotomists, and admins — are the secret sauce to providing great care. We strive to supercharge our providers with best in class tools. Importantly, we are committed to assisting, not automating, as we know that the human connection is a crucial ingredient in providing care. We are building tools to minimize time in front of the computer by accelerating input. We can summarize historical content and surfacing context-based suggestions. Rather than spending precious minutes chart-diving, we’re surfacing the most relevant information. We are at the bleeding edge of speech-to-text and ambient audio capture, to maximize facetime with the patient.

Opportunities:

  • NLP to understand provider intent for better autocomplete
  • Speech-to-text, speaker recognition, ambient audio capture
  • Clinical decision support
  • Recommendation engines for labs, meds, and specialty care
  • Search and discovery of clinical information

Population Health

When we zoom out and consider our patient population as a whole, we see a new set of opportunities. Which patients would benefit from various cancer screenings? Or smoking cessation? Or mental health interventions? To answer these questions requires big data and sophisticated patient modeling,

Opportunities:

  • NLP for social determinants
  • Patient health modeling
  • Intervention impact modeling

Infrastructure and Interoperability

Last but not least, we must acknowledge the elephant in the room: healthcare data is a mess. Providing high quality care requires the full story of the patient’s healthcare journey. That requires collecting health records from other institutions. While interoperability standards such as HL7 and FHIR continue to gain traction, faxes and paperwork are still the most common method of transmission.

Opportunities:

  • Document classification
  • OCR of semi structured documents
  • Data extraction from unstructured text
  • HIE data exchange with external partners

Are you an engineer who wants to make a big impact in healthcare? Are you a technology vendor that has solved one of these problems? We want to hear from you.

Size of healthcare:

https://www.cms.gov/research-statistics-data-and-systems/statistics-trends-and-reports/nationalhealthexpenddata/nationalhealthaccountshistorical.html

Artificial intelligence in pharma, health care: at the crossroads of hype and reality

https://www.statnews.com/2018/12/06/artificial-intelligence-pharma-health-care/

IBM Has a Watson Dilemma

https://www.wsj.com/articles/ibm-bet-billions-that-watson-could-improve-cancer-treatment-it-hasnt-worked-1533961147

Authors

Jonathan Hilgart is the Engineering Manager for the Machine Learning team which focuses on applying natural language processing to One Medical’s in-house health record.

Cody Ebberson is the Director of Engineering for the Data Department at One Medical which focuses on supporting interoperability between partners as well as storing and securing data throughout the organization.

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store