Physician's AI review caught 85% LAD narrowing missed by stress test

DoctorTro | Metabolic & Obesity Physician@DoctorTro

A patient with ongoing symptoms was told everything was fine.

His symptoms weren’t the classic presentation of heart disease. A cardiologist ordered a nuclear stress test. It came back negative, and he was reassured.

But he still had symptoms.

When he came to see me, I gathered his old records and imaging. I used an AI imaging tool to help quantify vascular and coronary calcification, liver fat, and bone density. No new tests needed - just making use of existing data…

His history revealed something that should have mattered a great deal: known coronary disease involving all three major vessels.

That matters because disease affecting multiple coronary arteries can sometimes produce a falsely reassuring nuclear stress test. When blood flow is reduced across multiple regions, the relative differences the test looks for can be harder to detect.

A negative test does not erase the patient’s symptoms or medical history.

Given his persistent symptoms and my level of concern, I ordered a coronary CT angiogram.

He went to a local hospital that overcharged his insurance and failed to complete the test ordered… (AI-augmented fractional flow reserve from the CT scan)

Basically they said “Whoops but thanks for the thousands of dollars…”

Again, the reading radiologist and cardiologist said everything was fine.

I reviewed the images myself because that reassurance did not fit the clinical picture. I saw an area concerning for significant coronary narrowing, with poor contrast opacification beyond the lesion.

I sent the study for additional AI-assisted plaque analysis and CT-derived fractional flow reserve, which helps estimate whether a narrowing is restricting blood flow. I could see it, I bet my 10 year old son could see it…

The analysis reported approximately 85% narrowing involving a long segment of the LAD, one of the major arteries supplying the heart. The flow assessment suggested a potentially significant restriction, warranting further evaluation, including whether a stent would be appropriate.

This patient had persistent symptoms, known coronary disease, and findings that deserved more than reassurance.

Without digging, without persistence and without technology this patient likely would have a cardiac event (10 year risk approaching 50%)

I am beyond frustrated with our medical system.

Patients depend on us to connect their symptoms, their history, their prior records, and their imaging. They depend on us to recognize the limitations of a test. They depend on us to keep asking questions when the pieces do not fit.

AI can help us examine information more closely. The responsibility to listen, investigate, and follow through still belongs to us.

How many patients hear “everything is fine” and go home still feeling that something is wrong?

It shouldn’t be this hard to deliver good care.