Nov 21, 2025, 3:21 PM ETConference Presentation, Fireside Chat, Panel

What Does Smart Implementation of AI in Health Look Like? | Future of Health Summit 2025

CMS manages a healthcare payment volume of approximately $2 trillion annually, serving as the largest payer in the global economy.
The federal government has identified $1.5 billion in potential waste within a six-state Medicaid pilot and over $1 million in a single Medicare state.
A new CMS Interoperability Framework was launched to allow patients to request medical records via a single app using modern identity verification (e.g., ID.me, Clear), eliminating reliance on physical CDs or faxing.
CMS is consolidating data from eight separate databases to create a unified, authoritative provider directory to eliminate redundancy in the private sector.
The agency is recruiting senior tech talent from major firms (Apple, Google, Microsoft, Amazon) and startups, shifting the internal engineering ratio from 1:275 (employee to contractor) toward a more robust in-house workforce.

CMS Administrator (Amy) spent 487 days (1 year and 3 months) obtaining a diagnosis for her daughter's juvenile dermatomyositis, a delay that worsened the patient's condition.
Her daughter now manages care across six health systems in four states, utilizing 21 daily pills and two monthly infusions, illustrating the fragmentation of patient data.
An AI analysis of her daughter's records corrected a misdiagnosis of ulcerative colitis (which excluded her from a CAR-T trial) to microscopic lymphocytic colitis, rendering her eligible for curative treatment.
In a separate instance, a conversational AI prompted a patient to recognize she had failed to taper high-dose steroids, preventing an unnecessary and costly ER visit.
Eric's pivot to healthcare was driven by a 10-year timeline of his wife's breast cancer treatment, during which he felt the medical system was technologically outdated compared to other industries.

US healthcare spending reached $5.7 trillion in 2023, growing at a rate of 7.5% annually, which threatens to exceed the size of the entire US economy within 20 years.
Eliminating all profits for drug manufacturers would save only $112 billion (approx. 2.5% of total costs), indicating that the primary driver of expense is systemic complexity and waste, not just profit margins.
Healthcare employs 22 million people, representing 1 in every 11 US workers, making structural overhaul difficult without technological infusion.
US life expectancy stands at 78 years, lagging behind countries that spend significantly less per capita (1/10th to 1/20th of US spending).
Healthcare-focused AI is projected to grow at 33% to 37% annually over the next several years.

Tempus currently stores 408 petabytes of data, connecting to 5,000 US providers to contextualize diagnostics in real time.
A Stanford University study cited indicated a 60% agreement rate among pathologists on diagnoses, highlighting the potential for AI to reduce diagnostic variance.
A Microsoft AI case study across 304 cases demonstrated that AI achieved 4x the accuracy of human physicians.
Approximately 3% of patients receiving a normal ECG result will suffer a heart attack or stroke within a year; AI-driven FDA-approved algorithms now predict undiagnosed atrial fibrillation and low ejection fraction.
Treating a single heart attack or stroke costs approximately $125,000, creating a strong financial incentive for AI-based preventative detection.

CMS aims to launch an initial version of a personal AI health companion within one year, designed to guide patients through care decisions and healthy behaviors.
The administration envisions a "Health Tech Ecosystem" where the government sets infrastructure guardrails (identity, directories) while the private sector builds competitive patient-facing applications (e.g., "Kill the Clipboard").
The long-term goal is to leverage AI to turn "good doctors into superhuman doctors" and prevent the healthcare system from becoming a $10 trillion unfunded liability by 2034.
Eric Tempus projects that the current administration's willingness to partner with industry and take risks represents a "defining moment" for healthcare innovation.
Both panelists emphasize that the primary barrier to adoption is not physician resistance, but rather the integration of technology into existing workflows (e.g., Epic, Cerner) without disrupting clinical time.
A major risk identified is the potential for patients to confront physicians with AI-generated data they distrust, requiring the simultaneous arming of providers with the same AI insights.