Imagine having a clear understanding of all the variables at play every time you have to make a big life decision. Things would be much simpler, wouldn’t they?

This would be particularly beneficial when it comes to physical wellbeing, especially since choosing to avail of healthcare is a problem for many people in Southeast Asia because of the associated risks and costs.

Even those who could afford the best treatment and amenities sometimes find it prudent to opt for a cheaper procedure. Ultimately, it comes down to individual preferences.

Unfortunately, that’s where the issue of unknown unknowns comes in.

“It’s very challenging to assess the risk and cost for individuals,” says Neal Liu, founder and CTO of uCare AI, a startup that helps reduce medical costs and maximize efficiency.

Neal Liu, founder and CTO at uCare AI / Photo credit: uCare AI

Reducing uncertainties in healthcare

According to Liu, while large healthcare organizations have access to population-based data that enables them to put together sound overarching guidelines that shape healthcare recommendations, things become much more complex at the individual consumer level. “There are all the health policies that try to help manage the cost,” he says. “But there are always outliers. How can we make sure that all the processes and policies are able to include them?”

For uCare AI, the answer lies in building predictive analytics models that are able to solve these challenges in a more comprehensive manner.

Specifically, the company is keen on using machine learning models and AI to forecast the future. “Given all the historical data, how can we look forward and make predictions from the risk and cost angles?” Liu points out.

Quality and quantity of data are fuel for effective AI development, and it’s no different in uCare AI’s case. In 2018, the startup established a partnership with Parkway Pantai, one of Asia’s largest integrated private healthcare providers that, prior to working with uCare AI, used to generate bill estimates through traditional statistical models. Because it was expensive to keep refreshing these estimates, they were updated infrequently, resulting in higher error rates.

Using the wealth of patient and medical data Parkway has, uCare AI built a predictive model to determine specific treatment options and the related costs. “If I get admitted to a hospital, the model will take all the various inputs from my unique profile to recommend a specific treatment and cost to me,” Liu explains. “This is better than just knowing the average cost for a typical male of my age, and I can make a much more informed decision.”

Parkway deployed this cost prediction solution to all four of its Singapore hospitals in November 2018. Since then, it has performed at an average aggregate accuracy of 82%. Confident in this prediction model, Parkway launched a price guarantee program in December 2019, which ensures that patients will be charged the initial price quoted by the cost predictor, even if additional treatments are added later in the process. This helps patients get a clear and early view of the finances involved.

It’s something that Liu is “super jazzed about,” because “it’s a huge step forward, and if we can start managing these risks and costs in a more transparent way, you could even buy insurance on the spot in the future.”

Using AI to help people make smarter decisions

The founder is thinking beyond just medical care, however. The goal is to use AI to help people make better lifestyle choices and have them think beyond the short term with more clarity.

“When I was young, I would eat a large pizza and wash it down with a huge bottle of soda,” the 50-year-old Liu says. “You know how bad that is for you, right? But with the immediate satisfaction I got out of it, I was like, ‘Hey, I’m good!’ But there are obviously consequences down the stream.”

The key lies in making AI accessible and transparent. Accessibility helps lower the barrier of entry, bringing in more AI talent into a market starved for them. Transparency, meanwhile, helps people from outside the AI field understand how it works, encouraging adoption rates.

Additionally, the data upon which models are built have to be both securely handled – with the appropriate security processes in place – and easy to train.

“We make it a point to have a human in the loop,” Liu adds. “We always seek the domain experts’ advice to make sure that we aren’t shortchanging ourselves in terms of how we’re looking at the data.”

UCare AI is bringing these principles together with AlgoBox, a cloud-based platform that allows users to easily train, develop, and deploy new AI models.

“AlgoBox works by establishing a baseline of machine learning operations,” Liu explains. “It can deploy production-ready models within three months, end-to-end, with 99.9% uptime.” In order to keep the model flexible even with drastic changes in parameters, the platform also enables continuous monitoring and learning of AI models in real-world deployment.

The AlgoBox platform / Photo credit: uCare AI

Liu brings up the Covid-19 pandemic as an example. “Cost and risk models are drastically different due to changes in treatment protocols,” he says. “Previously, a patient with symptoms of fever, cough, or runny nose would go through a simple outpatient visit. With the pandemic, doctors now need to administer more tests and monitor a patient closely to determine if there is a Covid-19 infection or not, which drives up costs.”

With AlgoBox, a data scientist can refresh and retrain the deployed models as they observe these data and model drifts to make sure they continue to work and behave as they are expected to.

Leaving a social impact

Looking ahead, uCare AI is collaborating with more partners to expand to other industries, starting with health insurance, which is a natural extension to its work in healthcare. Working with different stakeholders across the ecosystem and having access to their data stores also help the startup improve its model.

At a fundamental level, however, Liu is driven to use technology to make a positive difference in people’s lives. As the self-styled technologist puts it, “Only two things matter when you’re older: How much money you have for your retirement, and whether you’re healthy enough to enjoy it.” This motivated him to enter the healthcare industry, even with all the restrictions and challenges the field faces with technology and data.

“Machine learning and AI are so disruptive, we are only seeing the tip of the iceberg,” he says. “I would love to get more people excited about them because of the potential impact this technology can make on mankind. I want to bring together like-minded people and let them know that we can do wonders here.”

AlgoBox is a cloud-based platform built by uCare AI that allows users to easily train, develop, and deploy new AI models, simplifying the entire life cycle of machine learning with just a few clicks.

Find out more about AlgoBox and how it can accelerate AI adoption on its website.

This content was produced by Tech in Asia Studios, which connects brands with Asia’s tech community. Learn more about partnering with Tech in Asia Studios.