Healthcare systems must accelerate AI adoption in telehealth, says Sheba Impact CEO
Healthcare providers are moving too cautiously with artificial intelligence, missing opportunities to improve care delivery and expand access to patients who lack it, according to Avner Halperin, CEO of Sheba Impact, the commercialization arm of Sheba Medical Center.
The core problem is straightforward: aging populations cannot be served by bringing more people into hospitals. Telehealth, paired with AI, offers a practical solution to extend care beyond traditional clinical settings.
AI handles the work clinicians don't need to do
AI systems are already handling tasks that consume clinician time without requiring medical judgment. Conversational AI platforms contact patients before appointments, gather intake information, and resolve routine questions-work that previously fell to administrative staff or doctors themselves.
The efficiency gains are significant. In psychiatric settings, AI can complete 90% of intake work, reducing a psychiatrist's time from 90 minutes to 10 minutes per patient. The clinician then handles the remaining 10% that requires expertise.
AI also prioritizes which patients need immediate care and which can wait, addressing what Halperin calls "the most rate-limiting factor in the quality of care that all our systems provide"-getting patients in front of qualified clinicians.
Recent regulatory approvals signal momentum
Two recent approvals demonstrate growing acceptance of AI in clinical workflows. Israel's Ministry of Health approved Mentaily's AI psychiatric triage system, which uses an interactive avatar to assess patients. TytoCare secured FDA De Novo classification for an AI tool that analyzes eardrum images, establishing a new regulatory category in the process.
These aren't isolated developments. Telehealth adoption follows a pattern: it spikes during crises, dips afterward, but the overall trend moves upward.
The human-AI partnership works better than either alone
Halperin argues the evidence is clear: AI and humans together outperform either working separately. Physicians must make final decisions on major medical interventions. But the combination reduces errors and improves outcomes.
Healthcare's caution toward AI mistakes differs markedly from other sectors. Autonomous vehicles are at least seven times safer than human drivers, yet one accident generates headlines and reduces adoption. The same asymmetry applies to medicine: healthcare systems tolerate human error more readily than AI error, even when the data shows AI performs better.
"We're underusing AI," Halperin said. "We should actually be demanding that wherever possible, AI does anywhere between 10% to 90% of the work and the doctors do the rest."
Access to care becomes the real opportunity
The broader implication extends beyond efficiency. AI-enabled telehealth can deliver high-quality initial diagnosis to anyone with a smartphone, regardless of location. This closes gaps between healthcare in major cities and remote areas, between wealthy countries and developing ones.
That's where the practical value emerges: not in replacing clinicians, but in making expert assessment available to populations that currently have none.
Learn more about AI for Healthcare and how organizations are implementing these systems in clinical practice.
Your membership also unlocks:
