More and more healthcare startups bet that artificial intelligence will help them do more with fewer patients, such as caring for fewer patients.
The interests are high. A shortage of clinicians, severe margins and doctor burnout are putting pressure on healthcare providers.
The startup says AI could be a breakthrough that will help clinicians increase caseloads and earn extra cash without sacrificing the quality of care.
“The long-term vision is to continue to strip the facet of face care and deliver the care itself through technology,” said Daniel Perez, co-founder and CEO of new public physiotherapy company Hinge Health in an interview in May with BI.
But while some companies are leaning farther to automate more and more care, others are wary of patients' experiences of keeping their technology arm-length and handing over safety risks and clinician work to AI.
Instead, they are pushing AI behind the scenes to automate only non-clinical tasks. For some, that approach offers new leases to the lives of startups.
“Companies that have historically not reached profitability may have had to be able to go out of business, maintain and maintain their long-term impact,” said Stephen Smith, co-founder and CEO of NOCD.
More patients, more money
Some of the biggest AI movements in healthcare are happening in musculoskeletal care. There, companies like Hinge Health and Sword Health will actively use the technology to help physiotherapists become more efficient.
Related Stories
Business Insider tells the innovative stories you want to know
Business Insider tells the innovative stories you want to know
Sword Health added more AI applications for clinicians in 2024. Physiotherapists were expected to manage 200-300 patients at a certain time at the beginning of the year, but the company wanted these providers to manage an average of 700 patients by the end of 2024, BI reported in November. Sword declined to comment on the story.
Sword's efforts to that goal included using AI for clinicians to send to patients with each BI report. The company also said the AI helped physical therapists prioritize patients who may require more attention.
Hinge Health takes a similar approach in AI implementations, including patient provider messaging. Hinge Health refused to share the provider's caseload for this story, but the company says it used AI to reduce 95% of clinician time spent on physical therapy.
Hinge Health uses computer vision technology to guide patients with PT sessions at home. Following the IPO in May, the company said it is actively working on more AI applications to further automate care delivery, and has a large R&D team focused on its goals.
“At some point, whether it's 10, 50 or 200 years in the future, care delivery is automated with technology. That's a good thing,” Hinge Health CEO Daniel Perez told BI in May.
Daniel Perez, co-founder and CEO of Hinge Health Hinge Health
Virtual Pediatric Care Startup Summer Health is working to use AI to multiply the number of patients a provider can care for. Founder and CEO Ellen Dasilva doesn't think AI will replace doctors, but she said the startup is working on automating some clinical and administrator tasks.
Healthcare businesses that implemented minimal technology might expect providers to see 5-7 patients per hour, but Summer Health's mostly text-mesung-based care model allowed providers to see around 10 patients in an hour, Dasilva said. And with AI, “that number can be very easily doubled,” she said. “Our providers still have a lot of room to run.”
That may be particularly good news for specialties like pediatric care, where clinicians are short on. Dasilva said AI can help increase the number of remaining clinicians and help more patients get care.
Balance between operations and results
As some healthcare startups move on to everything with AI, others aren't selling much about the potential to improve care.
Omada Health, a chronic care company released in May, has long used AI behind the scenes, as clinicians suggest educational videos from Omada's content library to send to patients. However, it was slow to bring AI to patients and waited until the company could point out significant LLM improvements and train the model with its own clinical data.
The company released its first patient AI tool in early May. Answer member nutrition questions. But Omada Health co-founder and CEO Sean Duffy said the company is not intending to significantly replace patient provider interactions with AI across its business.
“I haven't yet found anyone who feels responsible for chatting with GPT,” Duffy told BI at Omada's IPO in May. “Always cite the reason why artificial intelligence is a buzzword and artificial empathy isn't. You can expect to have a component of aggressive people all the time.”
Virta Health, a diabetes care peer at Omada, has been using machine learning since 2017, according to founder and CEO Sami Inkinen. Between 2015 and 2020, Virta was changing negative gross profits as he was thinking about ways to use technology such as machine learning to achieve better results. But generative AI helped startups get that number on the green. Today, Virta boasts a total margin of around 60%, according to Inkinen.
Virta injected AI into many parts of the business, including patient interactions, with personalized care plans and chatbots powered by AI, helping patients answer nutrition questions 24/7. However, Inkinen said that AI is not making clinical decisions and will not replace Virta's business provider.
He added that Virta has no target for provider caseloads, and that the company is rigorously focused on providing better patient outcomes and improving its finances along the way.
Sami Inkinen, co-founder and CEO of Virta Health. Virta Health
Mental health startup NOCD uses AI to tackle a wide range of management tasks, including clinical note oralization and revenue cycle management. CEO Stephen Smith said the technology helped the company reach profitability, allowing NOCD to grow its therapist network while keeping operating costs low.
Startups have a chatbot called Robin, which allows therapists to ask for information during sessions. However, AI does not interact directly with patients and does not intervene unless the clinician seeks its help.
Smith said he was careful about his focus on patients with obsessive-compulsive disorder. This is usually classified as a serious mental illness. NOCD may be able to increase provider caseloads with AI, but he said it is ultimately not a short-term goal for the company.
“I know that all these different projects deal with tough groups of people, so we have to go with caution,” Smith said. “Turning AI into clinical delivery of care requires a lot of research, mainly because we want to make sure it is super, very effective and can create a safe experience for our members and therapists.”
