Could AI be the answer
Are we running a blood‑letting empire in medicine—and could AI be our salvation?
Every month, thousands of patients get the familiar ping: “Please book your blood test.” Annual reviews, quarterly checks, dose‑change panels… the cycle never ends. Behind the scenes, an entire workforce orchestrates reminders, bookings, form‑filling, car‑parking logistics, carbon emissions and, of course, mountains of paperwork. By the time the results land, nearly 100% are normal.
Here’s the kicker: after 25 years of routine DMARD and medication monitoring, I can count on one hand the routine checks that flagged anything actionable. (I’m not talking oncology protocols or acutely unwell patients—those are a different ballgame.)
First, do no harm… but what about the stress, clinic bottlenecks and the “blood‑test fatigue” our patients feel? Patients spend more time coordinating their phlebotomy than managing their symptoms.
Could AI slice through the tedium?
Frontiers in Medical Engineering highlights how scalable AI models can sift hidden patterns in routine labs—beyond what classic decision‑support ever spotted—and drive smarter, leaner monitoring protocols (Frontiers).
An ICU study showed machine‑learning algorithms predicting which future blood tests would actually change clinical management—potentially cutting unnecessary labs by up to 25% without compromising safety (PMC).
Imagine a world where:
Young, fit patients on stable regimens get 6‑monthly rather than 3‑monthly tests;
AI flags only those with a high probability of an abnormal result;
Lab teams focus on the critical 5%, not the 95% of normals.
A Provocative Thought
Are we over‑monitoring stable patients out of ritual rather than risk? If you’d trust an AI‑powered “blood‑test concierge” to decide your next phlebotomy slot, would you cut your visits in half?
Further Reading
Santos‑Silva et al. “Artificial intelligence in routine blood tests” (Frontiers)
Lee et al. “Predicting information gain in ICU lab testing” (PMC)
Over to you!
Have you ever had a routine test uncover something critical?
Would you feel safe if an AI model told you, “Not due for bloods till December”?
How could we pilot AI‑driven lab protocols in your clinic?