Top

‘We’re improving patient safety’: How one CS grad is using AI to personalize health-care operations

Alumnus Devin Singh is the co-founder of Hero AI, a clinical automation software platform that uses AI to improve health care for both patients and providers. (Photo: All photos courtesy of Devin Singh)

Computer science alumnus Devin Singh (MSc 2020) is on a personal mission to transform health care by leveraging artificial intelligence.  

Through Hero AI, a health tech company he co-founded that launched in June 2020, Singh is urgently trying to tackle ongoing complex challenges in health systems, including high patient volumes and limited resources.  

The fully encrypted, PHIPA-compliant clinical automation software platform can take streaming electronic health record data, then apply decision logic and AI modelling in order to present insights in a web app, dashboard and patient- or provider-facing mobile apps. Ultimately, it sets out to improve the patient experience and provide customizable solutions for hospitals.   

“We’re able to look to the hospital and say, ‘What are the problems or pain points that you’re experiencing in your workflow?’ Whether it’s an emergency department or the wards, they’ll give us their top 10 issues, then we model out automation solutions, and deploy those insights to people on the ground,” explains Singh, who is a staff physician and the lead for clinical AI and machine learning in paediatric emergency medicine at Toronto’s Hospital for Sick Children (SickKids). 

Singh says the technical innovation on the platform’s back end enables Hero AI to rapidly build solutions in a matter of hours, rather than months or years it would typically take. 

“It allows us to go into a hospital, and say, ‘Here’s an AI-powered automation platform at a baseline, but our expectation is that you will rapidly customize this to meet your specific needs,” he adds. 

The Hero AI platform provides health-care teams with AI-powered dashboards that show emergency department activity, including patient census, number of patients waiting to be seen, and other custom metrics.

Driven by a case when he was a medical trainee, Singh recalls the effort trying to resuscitate a patient who ultimately died. He says the experience was devastating, but one that “shook me and lit a fire in me.” He knew he wanted to innovate and find ways to prevent future adverse patient outcomes. Upon reflection, he began to realize AI and machine learning could offer solutions to the complexities and challenges in health systems, like prolonged wait times.  

“If you look at my research career, my industry career, my clinical career all combined, you’ll see that if you take that case, and that pipeline that happened that led to that terrible outcome, I am tackling each of the failure points along that pipeline, to make sure no patient ever goes through this ever again,” Singh says. “That’s why I’m building this, I really have that case at heart and all of the research I do is really about preventing that from happening.” 

Setting out on this journey, he taught himself how to code and build machine learning models online and pursued his master’s degree in computer science at the University of Toronto.  

Singh credits his U of T education with directly influencing the work he’s doing at Hero AI. He was co-supervised by professors Michael Brudno and Anna Goldenberg, and met classmate Pouria Mashouri through the program, who would later become one of Hero AI’s co-founders and its chief technical officer.  

“Besides just raising my own technical capabilities, the master’s of computer science at U of T really helped me meet the right people to then catalyze the creation of Hero AI,” says Singh, who is also an assistant professor in U of T’s Department of Paediatrics and cross-appointed to the Department of Computer Science.  

He notes Hero AI immediately enables hospitals to increase capacity and improve patient safety, which in turn allows for the deployment of scarce resources “in a much more targeted way, further improving patient safety and improving wait times.”  

“It really adds a level of agility to a hospital’s ability to respond to different emerging issues and crises,” Singh argues. 

Singh says he also understands why there is hesitancy to adopt AI in health care, especially given privacy concerns. He notes this is top of mind for the team at Hero AI, who has done extensive stakeholder consultations to ensure they are meeting patient expectations, as well as complying with privacy and regulatory landscapes.  

“When we talk to patients, they love the idea of being seen faster. They love the idea of better situational awareness around their care while they’re in hospital. But they want to know what is being powered by AI and what is being powered by a human. They just want that transparency.”  

Any decisions, recommendations, or communications initiated by AI algorithms are clearly explained to patients that it’s coming from an algorithm and not a human provider, Singh adds, explaining it’s crucial for patients to know when and how AI is assisting. 

Understanding how to deliver that transparency as Hero AI prepares to scale up, is part of the company’s ongoing work. By interviewing patients in the SickKids waiting room, they are collecting feedback on how to provide transparency that is “truly nuanced and genuine.” This is especially key for the company as they look to deploy future clinical automation tools that can make predictions about testing needs, Singh notes. 

With Hero AI already in use at SickKids and IWK Health Centre in Halifax, Singh says he’s seeing firsthand the impact on both providers and patients. He notes how the tool can help reduce emergency department wait times for patients experiencing mental health crises.  

“With that streaming data, Hero AI identifies that a patient is in a mental health crisis, and immediately automates the notification to the psychiatry team in near real-time and allows patients to receive the psychiatric care and support they need before they’ve even been seen by the emergency doctor. As a result, when the psychiatry team receives an automated alert from Hero AI, patients who are having a mental health crisis get seen by psychiatry much faster than before. And that’s a huge innovation in mental health care that’s happening in the emergency department at SickKids,” says Singh.  

Monitoring wait times also applies to patients who are experiencing acute surgical concerns, like testicular torsion, Singh explains. Hero AI can identify if a high-risk patient is waiting too long and can notify a health-care provider to see them, especially during times when the hospital is experiencing high volumes.  

The Hero AI platform uses AI algorithms to automate the detection of high-risk patients and notify a clinician.

As health systems continue to face rising demands for care, Singh explains prolonged periods of waiting create more risk for patients and limited human resources need to be deployed in “intelligent ways.” 

“We’re improving patient safety, we’re improving the deployment of scarce health-care resources, through process automation and through AI-powered insights.” 

The Hero AI platform provides a personalized patient dashboard showing wait times, position in queue and how busy the emergency department is.

Singh describes Hero AI as technology that is “sitting beside the patient” and “advocating loudly” through process automation to ensure high-risk patients “get seen as fast as they can.” He notes that through past notes and patient history, Hero AI can detect when a family comes into the emergency department with a child who has autism, and it will begin advocating to the child life specialist and to the social worker to make sure the care the child receives is adapted and helps them cope with that emergency department experience. 

“An institution that has Hero AI deployed means that you’re going to have an elevated patient safety experience,” Singh says. “We’re going to make sure the highest risk patients aren’t left waiting too long.” 

As Singh and the team at Hero AI are ready to scale and deploy to other sites, his hope is for Hero AI to be used nationwide.  

“Hero AI is going to be really focused on ensuring literacy around responsible AI across Canada,” Singh says. “In doing so, we’ll be able to not only help Hero AI scale across the country, but other really talented local tech companies who have great solutions will hopefully then get better market penetration in our Canadian landscape as well, because there's a lot of things to improve in health care. Hero AI wants to play its part, both on the AI literacy piece, but also on the tech innovation.”