Marco's Morning Report

Friday, June 26, 2026 · Cape Coral, FL

Morning, Marco — hope the floor was merciful. Clinical wires are quiet again (no fresh guideline drops in 48h), so today leans home: a real water emergency in Puerto Rico, the Mets officially flipping into sell mode, a Florida Medicaid fight on your doorstep, and a debate about whether the AI on your phone now beats the FDA-cleared stuff. Coffee, then bed.

Section A — Top Stories

Confirmed-fresh, last 48h. What actually matters today.
Puerto Rico

Water crisis hits PR — National Guard activated as families go days without service

Gov. Jenniffer González has activated the National Guard and emergency crews as thousands across the island face severe water shortages — some homes dry for up to two weeks — just as summer heat advisories begin. Guard trucks (four units, 2,000 gallons each) are running distribution. It lands on top of the chronic grid problems, so people are losing power and water at once.

Why it's on your radar: family and diaspora ties, and it's a textbook heat-plus-dehydration setup — exactly the elderly profile you admit. If relatives there are older or on diuretics/RAAS agents, a check-in call about hydration and heat is worth it. ABC News

Mets

Update: Mets flip to sellers — Peterson traded to the Cubs, Semien to the IL

The trade chatter you've been tracking became real. New York dealt LHP David Peterson (3–6, 6.09) to Chicago for 1B/DH prospect Cole Mathis (Cubs' No. 13). Hoyer said Cubs rotation injuries (Cabrera, Brown to the IL) forced the move. Meanwhile Marcus Semien went on the 10-day IL with a left hip flexor strain and Ronny Mauricio was recalled. After a brutal six-error doubleheader loss and a five-game skid, Passan reports rival execs expect "plenty more" — Peterson looks like the first domino, not the last.

Talking point: this is a soft white-flag. Watch whether Stearns moves a bigger chip next; Lindor and Soto are reported untouchable. MLB.com · amNY

Florida

Florida's largest pediatric group challenges AHCA over a Medicaid rate cut

Pediatric Associates — which serves ~300,000 Florida kids — filed a challenge with the Division of Administrative Hearings, alleging AHCA improperly changed Medicaid reimbursement rates for 2025–26 and warning it may have to drop Medicaid services if the 2026–27 rates aren't corrected. It runs parallel to the ~$8B in supplemental hospital Medicaid payments CMS approved this spring.

Why it matters to you: Medicaid rate-setting in Florida is the same lever that shapes inpatient and post-discharge access for your panel. A pediatric provider threatening to pull out is the canary — worth half an eye as the 2026–27 rate cycle plays out. WUSF Health News Florida

Section B — The Debate

One live argument, both sides, named voices. No verdict.

Should clinicians lean on general-purpose AI at the point of care — or stick to FDA-cleared clinical tools?

What lit it up: a June 23 Nature Medicine head-to-head found general-purpose frontier models (GPT-5.2, Gemini 3.1 Pro, Claude Opus 4.6) beat both cleared clinical tools — OpenEvidence and Wolters Kluwer's UpToDate Expert AI — across every benchmark, including the "real clinical query" set drawn from live point-of-care questions. The catch: the cleared tools went through an FDA pathway; the general models that beat them did not.

For — let clinicians use the better tool

The benchmark exposes a validation gap, not a safety win for clearance. As the Clinical Trial Vanguard analysis argues, an FDA stamp here certifies a process, not point-of-care performance — and on the messy, unstructured questions that actually arrive at the bedside, the general models answered better. Blocking the stronger tool in the name of a clearance that doesn't measure real-world accuracy, the case goes, protects the regulatory pathway more than the patient. The honest response is to validate and govern the models clinicians are already reaching for, not pretend they're inferior.

Against — benchmarks aren't bedside safety

Winning a benchmark isn't the same as being safe in a live encounter, and the uncleared models carry no accountability trail when they're wrong. Writing in Physicians Weekly, clinicians stress that diagnosis, risk stratification, and prognostication hinge on patient-specific context generative models can't fully hold — a human has to review every output before it touches care. A June 22 ethics piece by Cabrera and colleagues warns that general LLMs' answers swing with how the prompt is phrased, a fragility that's tolerable in a chatbot and dangerous at the bedside. Clearance is a floor, not a ceiling — and there isn't one yet for these tools.

Where it might land: no regulatory pathway yet calibrates general-purpose models for clinical decision support, so expect the near-term answer to be "use as a consulted second opinion, with a clinician signing off" rather than a green light to act on raw output.

Section C — On My Radar

Noteworthy, not urgent.

Section D — Trends to Watch

Signals, not headlines.

Section E — Ideas & Opportunities

Small, actionable.
Section H — Background / Already Covered
📚 Archive · last 7 morning reports