Now that two out of three physicians are using AI, who better to sit down with than healthcare expert Sara Gershfeld to dive into how it’s being used among practices. Gershfeld regularly writes about AI trends on Substack, and advises industry startups on the best ways to integrate AI into their behavioral health and care delivery organizations.
“I’m really interested in where technology and service delivery intersect,” says Gershfeld. “And I’ve always been really interested in innovations in technology and how organizations can use technology to improve operational efficiencies and improve quality outcomes.”
There’s a reason CEOs read our emails
Every few weeks we’ll send the most important info so you stay informed
Here are a few of Gershfeld’s current observations for the tech-curious, and beyond.
Don’t Just Rush Out and Buy Any AI
“It’s really important to understand what is true AI, and what actually exists beyond the hood,” says Gershfeld. “And then also to validate the quality of the technology being used.”
First, it’s crucial to know the difference between AI and automation—those terms are often interchanged. “Automation has existed for a very long time,” Gershfeld says. “Just because you have something that automatically does something, it doesn’t necessarily mean that it’s AI. AI can exist in different forms, and it’s important to understand what data is being used to train the artificial intelligence model.”
Which is why it’s so important to choose the right AI model for the task. “You definitely need to be cautious around what you’re implementing and make sure that the quality of the product is high,” says Gershfeld. “I would suggest that any individual who’s trying to determine whether to integrate AI should think of the function of the behavior. For example, if the function of taking manual notes is so that you can remember what the conversation was about later, and if there’s a tool that allows you to not have to take notes and still remember what the conversation was about, then it’s a function that exists. Ultimately, the goal should still align with what you’re trying to achieve.”
A Human Touch Is Essential
The biggest learning curve is mastering the unique training needs of an AI model. “It’s no different than becoming proficient in Excel or word processing,” says Gershfeld. “One thing you hear a lot about is prompt engineering proficiency. The AI is only as good as the prompting that you’re putting into the system. So anyone using AI has to be really smart about what type of questions they’re asking, how they’re asking, and what type of information they’re feeding in that broadens the perspective. Each AI platform has its own interface and capabilities.”
AI models need to be trained to work the way that fits your organization best.There are two common ways to do this: using data your team has already collected to teach the AI, or having humans guide the training.
“I think there are pros and cons to both of those,” says Gershfeld. “The other piece is knowing that it needs to evolve with additional data coming in, so it can’t be stagnant. It can’t be based on a model that was created a year or two years ago, when there’s new data that’s been pushed into the system.”
”78% of healthcare workers who implemented AI said it expedited note-taking
It Eases One of the Biggest Industry Stressors
In a recent Stanford study, 78% of healthcare workers who implemented AI said it expedited note-taking and freed up time to prioritize patient care. In an industry faced with staffing shortages—and with fewer staff and more paperwork for clinicians, keeping employees happy and finding quicker, better ways to handle daily tasks isn’t just a nice idea—it’s a must.
“One piece around AI is this idea of standardization,” says Gershfeld. “If something is manual, for example, there’s variability around how it’s done, and then what that quality looks like from clinician to clinician. But if there’s a system that can automate and think a little smarter around how things are done, then there’s this opportunity for better quality outcomes and more standardization.”
Gershfeld says that getting rid of excess pen and paper analysis reduces errors and employee stress, which leads to less burnout, better staff retention, and more internal resources focused on patient care and company growth. “There’s a really big opportunity to upskill your clinicians,” she says. “I think AI can build on that learning and provide them with tools.”
The Benefits are Undeniable
Gershfeld is seeing some incredible AI use results via Frontera, an ABA technology company that she advises. They created video AI known as “digital phenotyping” to evaluate patient behavior in ABA therapy sessions. “The model that they have can rapidly process video data to reduce the time needed to analyze it,” says Gershfeld. “Sometimes data analysis can take days or weeks, and this happens immediately.”
“There’s some companies that are using AI to do early detection and screening,” Gershfeld continues. “They are able to use AI to determine whether a child has autism a lot sooner than a clinician would be able to analyze it.”
“Sometimes, especially for younger clinicians, it’s more challenging to individualize,” says Gershfeld. “And you find that they might be copy and pasting or just reusing goals from prior clients. So AI can really help personalize that.” Gershfeld notes that NonBinary Solutions, a technology and consulting firm that designs systems for autism service providers, uses AI for similar real-time decision support. “A clinician plugs in information about the patient, and it’ll give decisions around what type of treatments and planning to do,” she says.
AI Is Here to Stay, So Embrace it Responsibly
A late 2024 McKinsey & Company survey of healthcare leaders reports that 85% of participants were researching, working on adopting, or had already integrated AI capabilities.
“There’s certainly a skepticism around whether AI will replace clinicians, and I think that AI is meant to enhance clinicians,” says Gershfeld. “I’ve heard this quote that I really like, which is, ‘AI is not going to replace the clinicians, but clinicians using AI will replace clinicians not using AI.’ I think that’s really powerful, because it’s just a tool to enhance what you’re already doing as opposed to something that can replace you as an individual.”
About Sara Gershfeld: Gershfeld—a key speaker in our recent Unlocking Growth in ABA webinar—has been a trusted voice in healthcare for years. She’s the founder and former CEO of ABA therapy accreditation pioneer Jade Health, and she established the Autism Investor Summit.
On the Subject of Streamlining
Apploi is an all-in-one employment platform that empowers you to hire, onboard, and schedule top talent 71% faster.
