Does your Doctor have Artificial Intelligence built in?  Healthcare 2.0 begins now.

Artificial Intelligence

Does your Doctor have Artificial Intelligence built in? Healthcare 2.0 begins now.

Tuomas Yla Kauttu

by: Tuomas Yla Kauttu

December 10, 2017

Dr. Watson, I presume? Augmenting healthcare with AI

Health is one of the most precious assets that humans have over their lives. Yet, it can be something that is overlooked for a long time and only given attention when the first symptoms of a disease arise. The general discussion on healthcare is shifting from sick care to preventive medicine where future illnesses are tackled years before they ever occur. This is done with more holistic treatment plans which aim to increase the overall wellness of people combining crucial factors such as diet, exercise, sleep, wellness and work-life balance together. Embedding exponential technologies including AI to healthcare will help to augment this patient experience and enable people to get more personalized service. (1)

Complementing a doctor’s human intuition with the precision and completeness of AI in medical diagnosis could be one of the greatest revolutions in healthcare since hand-washing. The unbelievable scale of research on human health is more than any single human brain can usefully process and AI is starting to out- perform even the best human doctors. UK-based startup Babylon Health is developing an AI doctor app which aims to empower doctors and healthcare professionals to be even more accessible to a larger share of the population. The company’s scientists believe that their technology will soon be better at diagnosing diseases and personal health problems than doctors but the company states that it’s more about collaboration than competition. (2)

There are currently hundreds of startups entering the healthcare industry with different types of AI solutions and established corporations and investors are increasingly funding this growth. Healthcare is the hottest area of AI investing with a total amount of 270 deals in the category since 2012. Machine learning and predictive analytics are being applied to an enormous amount of different use cases from virtual nurses, drug discovery and population health management to oncology and diagnosis from digital images. Routines tasks such as scheduling appointments and patient monitoring can be automated using AI as well as bring bring efficiency and scale to patient work as doctors are able to operate multiple machines and robots at the same time. Many of these new startups are significantly lowering the costs of providing healthcare services or augmenting the way these services are provided. The variety of different AI startup categories is depicted in figure 1 below. (3, 4)

Figure 1: 106 startups transforming healthcare with AI (CB Insights 2017)

AI has great potential in helping doctors improve their diagnoses, forecast how diseases will spread and personalize customer treatment plans. The technology is proficient in pattern recognizition from large volumes of data such as historical patient data, medical images and epidemiological statistics and can draw inferences by analyzing this data. A number of companies are using machine learning for tailored treatments such as optimizing rehabilitation of stroke patients, designing individual cancer-treatment regimens and recommending the best time for a patient to take medication based on their metabolism and other data points. (4, 5)

Hospitals can harness AI to enhance their capacity utilization with more efficient allocation of resources and automate routine tasks such as the process of entering a hospital or a routine doctor appointment. Virtual agents can serve as a main point of contact in several different use cases including in-hospital and remote consultations, diagnosis of basic illnesses, treatment of chronic diseases and prescribing drugs in the future. They can schedule tests for patients, register them into the hospital and help patients to navigate inside the hospital from bureaucracy to being on time for appointments. Several hospitals are already experimenting with machine learning applications and collaborating with AI technology providers. Mayo Clinic uses image recognition and machine learning in radiology to identify genetic abnormalities and cancer cells from MRI and X- Ray images reliably and quickly. Moorfields Eye Hospital in London has partnered with Google DeepMind and Cleveland Clinic Lerner College of Medicine is working together with IBM Watson to realize the potential of using machine learning. Different use cases of AI in healthcare are depicted in figure 2 below. (5, 6)

Figure 2: AI in healthcare (McKinsey 2017)

Insurance companies are also excited about the potential of AI as it enables better ways to predict patient behaviour and predict disease probabilities for different customers. Insurance companies will focus on incentivizing preventive healthcare, disease management and wellness plans and address patients unhealthy behaviour before they become patients. Insurance companies, healthcare providers and pharma companies will partner together to facilitate these pay-for-performance models and an interesting example of this is Discovery Health insurance company from South Africa which tracks it’s customer’s diet and fitness activity and rewards healthy behaviour with incentives. (5, 7)

Adopting AI in healthcare still has several bottlenecks including the availability of high-quality datasets in standardized formats. Patient data is fragmented to electronic medical records, lab results, doctor’s notes, health insurance claims etc. and merging all this data together from multiple sources into an integrated consistent database is very difficult. Other hurdles include existing cultural barriers, strict regulation of the healthcare industry and consumer behaviour regarding sensitive personal health data and privacy. Adopting AI technology needs a lot of in-house education for hospital staff and doctors need to better understand how the AI reaches it’s diagnosis or chooses a treatment plan as currently the knowledge of how deep learning neural networks reach their decisions is limited. (5, 7)

We are still decades away from fully automated hospital experiences where virtual nurses describe medication, robot doctors perform surgeries and machine learning algorithms determine the patients rehabilitation plan. Despite all the technological advancements made in AI, healthcare is fundamentally a human-centered interaction where human touch, conversation and empathy are key. Doctors’ and nurses’ ability to comfort patients in their moment of greatest despair is irreplaceable and hard to augment with technology. As Dr. Watson takes more responsibility in hospitals in the future, human employees have more time to focus on these precious moments and support patients on an emotional level when they need it the most.

Sources

(1) Singularity University (2017). Exponential Medicine. Retrieved at: https://exponential.singularityu.org/medicine/

(2) O’Hear S. (2017, April 25). Babylon Health raises further $60M to continue building out AI doctor app. Retrieved at: https://techcrunch.com/2017/04/25/babylon-health-raises-further-60m-to-continue-building-out-ai-doctor-app/

(3) CB Insights (2017, April). The State of Artificial Intelligence. Retrieved at: https://www.cbinsights.com/research-artificial-intelligence-trends-report

(4) CB Insights (2017, February 3). From Virtual Nurses To Drug Discovery: 106 Artificial Intelligence Startups In Healthcare. Retrieved at: https://www.cbinsights.com/research/artificial-intelligence-startups-healthcare/

(5) McKinsey Global Institute (2017, June). Artificial intelligence — the next digital frontier? Retrieved at: http://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/how-artificial-intelligence-can- deliver-real-value-to-companies

(6) Lomas N. (2016, July 5). DeepMind partners with NHS eye hospital to conduct AI research. Retrieved at: https://techcrunch.com/2016/07/05/deepmind- partners-with-nhs-eye-hospital-to-conduct-ai-research/

(7) Accenture (2017). Technology Vision 2017. Retrieved at: https://s3.amazonaws.com/assets.accenture.com/PDF/Accenture-Tech-VisionReport-2017.PDF