How Artificial Intelligence Will Make Healthcare Better (and Even More Personal) Than Ever

There’s an old joke – although maybe it’s not funny – about the quality of healthcare: Fifty percent of the doctors in the world finished in the bottom half of their class.

Clearly that’s harsh, but one thing is true: the nature of healthcare often requires doctors and medical professionals to make quick decisions based on limited sets of data.

Which makes healthcare ripe for improvement through artificial intelligence.

One company working to do that is Lumiata, a startup striving to improve individual quality of life and lower the cost of healthcare delivery by predicting health risks with data.

Intel Capital made it possible for me to speak with Lumiata CEO Ash Damle about how data – and connectivity – can help improve the quality and availability of outstanding healthcare.

You have all this data and all these data streams… but conceptually, where do you even start?

Our focus is building what you could think of as an AI (artificial intelligence) physician.

When we think about healthcare overall, healthcare runs on lists. At the end of the day, major healthcare operations that have impacts on quality of care and costs, boil down to a list that payers, doctors or care managers have to tackle. These lists are very expensive to produce, and are highly imprecise and ineffective in terms of taking action. We’re creating an operational AI that will impact the day-to-day activities of health operations – from risk management to quality reporting to care management. The goal is to leverage AI to provide insights for payers and their physicians by using data at a massive scale. The result is a bigger return on healthcare activities: earlier, more precise, and more impactful.

The goal isn’t to replace physicians or healthcare professionals, but give them better decision-making tools.

Absolutely. Imagine if you harness the power of all the data that is out there: research, labs, claims, social data… and enrich all that data with AI to predict, for each individual person, how his or her health will evolve over the next months or years – and, just as importantly, the chain of reasoning behind why those things might occur.

That analysis will allow prescriptive AI. Sometimes that might change how you treat an individual patient. Sometimes that might change how a healthcare system deploys its resources. The goal is to get the biggest return for your effort, and to give the consumer the biggest return on their healthcare spending.

Can you give me an example?

Imagine if you had the ability to more accurately predict who will become a diabetic unless you intervene now. Or imagine if you had the ability to predict a sub-set of current diabetics who you can -- because of some advancement in science and treatment -- better help now.

The key is to use all the data we have, and will have, to make the biggest clinical and financial difference. Making that difference is what every physician hopes to do – and we want to help them do just that.

Of course, but that’s hardly an overnight process.

We think about this in three stages. The first is using AI to collect, confidentially and always with permission, all the available data we can. The second is using AI to predict, and the third is to use AI to actually engage patients.

The first key is to take that raw data and convert it into a standard format… but also to dramatically enrich that data with medical knowledge.

Then we can start to predict what will happen. With deep learning, it will be possible to take where each person’s health is today and predict how it will change across different time frames. With improved AI, we can surface the key signals from all the noise.

After that comes engagement, and the clinical rationale: How do we rank order patient needs? How do we prioritize, as an organization, so we make the best use of our resources? Our goal is to show healthcare professionals who is at risk, and why they are at risk, and what to do about it.

A fear many people have of artificial intelligence is that it removes the human factor.

We think of data as a service. Plug that into health management activities, and the professionals can now make better decisions because they have immediate access to better and smarter data.

Unfortunately, it’s impossible to have warehouses full of physicians looking over data. So how can you provide thousands of physicians, especially as we start moving to real-time data, with the information they need? The only way to meet that demand is to automate with AI, at a massive scale, because that allows every person in the organization to deliver the greatest value. We’re not talking about replacing people – we’re talking about giving them tools to be even better than they already are.

You mentioned real-time data. How will you incorporate that? With 5G connectivity on the horizon, we’ll see a flood of data coming from all sorts of new devices and sensors.

That’s an important point: How do you get streaming processes to work well? Our relationship with Intel Capital has allowed us to plan for that. From an infrastructure point of view, we’re already structured to do just that.

All that data will help create a longitudinal record of each person, which will allow AI tools to monitor, evaluate, predict… The churn of real-time data will allow for real-time changes and interventions.

Combine the real time with the episodic, in a standardized way, understand what all those inputs mean… that will help organizations make maximum use of their resources. And, of course, the patient benefits.

When we think about the long term, we like to think about getting to a place where care is delivered to us before we even know we need it.

Delivering that level of care will certainly be dependent on the amount of data we can obtain in nearly real-time about each person. 5G should allow for richer inputs and richer data. For healthcare to be more predictive, greatly increased connectivity and data flow will be key.  When you can access extraordinarily rich information, you can understand a patient’s health in a completely new way.

That will be revolutionize how we deliver care to people.

Since one of the goals of using AI in healthcare is to make better predictions… as you look forward, what excites you about what you’re working on?

One of the things that excites me is to better understand what works for whom.

Currently physicians at times are forced to make a prediction based on a prediction. Without accurate, timely data, predicting based on a prediction creates a compounding error. With AI, if you can make sense of data to eliminate that first prediction, because you know, then your next decision can be a lot more accurate, even if it’s a prediction.

Physicians try to understand where you are, decide where they think your health is going, and then decide what they can do about it. Improving the quality of care, of risk assessments, of how organizations deploy their resources… all that will impact people’s lives, and that’s extremely exciting. That will be transformative.

We’re trying to move medicine and medical science to the next level, and that is hugely exciting.

We all want value-based healthcare, and this is an objective way to build a platform that helps us do just that. What AI can do for healthcare is profound, because it can help us lead the healthiest lives possible and enjoy the opportunity to live the lives we want.

That’s what excites me most.