𝑺𝒉𝒐𝒖𝒍𝒅 𝑨𝑰 𝒊𝒏𝒕𝒆𝒈𝒓𝒂𝒕𝒆 𝒘𝒊𝒕𝒉 𝑪𝒍𝒊𝒏𝒊𝒄𝒂𝒍 𝑾𝒐𝒓𝒌𝒇𝒍𝒐𝒘?
𝑰𝒏𝒄𝒓𝒆𝒂𝒔𝒊𝒏𝒈𝒍𝒚, 𝑵𝑶 ... 𝒂𝒏𝒅 𝒕𝒉𝒂𝒕'𝒔 𝒂 𝒈𝒐𝒐𝒅 𝒕𝒉𝒊𝒏𝒈!
This advice is becoming more questionable, if not obsolete.
Yes, AI scribes, apps like OpenEvidence, and use of AI in radiology, cardiovascular care, and serology have integrated into current workflows. They've compressed timeframes, improved diagnoses, and allowed HCPs to be more present with patients and reduced cognitive loads that have resulted in HCPs burnout.
However, clinical workflows are often built around the pysical design of legacy facilities, outdated scopes of practice, and changing locations as portrayed in my 5Bs of Healthcare Evolution framework. In addition, AI in Healthcare deployment is rapidly moving along the five levels of the PCESU scale.
These AI innovations don't fit with clinical workflows; they change them.
1️⃣ Ultrasounds usually require a specialized sonographer with years of training in probe manipulation. Caption Health (a GE Healthcare company) provides real-time, turn-by-turn "GPS" guidance. It tells a medical assistant or nurse exactly how to tilt, slide, and rotate the probe to capture a diagnostic-quality image. This replaces the need for a sonographer for routine scans. And, the scan can occur at an ER bedside or a retail pharmacy by a tech.
2️⃣ Anesthesia has been a procedure where an anesthesiologist must stay in the room (by law or regulation) to manually administer anesthesia drugs. AI systems (like McSleepy), developed at McGill University, monitor brain waves (EEG) and vitals to deliver the exact dose of propofol or remifentanil every few seconds during surgery. Today, an anesthesiologist can "supervise" up to 10 operating rooms simultaneously from a central cockpit.
3️⃣ Heart Failure can require surgery (implanting sensors) or frequent hospital visits for fluid-level checks by cardiology nurses. Now, AI-powered sensors (RFID or Optical AI) can detect lung fluid buildup and heart rate variability in a patient's home. Clinic visits are removed for routine monitoring and surgery, and the AI can predict a "crash" weeks in advance, alerting a remote monitoring center.
4️⃣ Diabetic Retinopathy Screening usually requires a referral to an Ophthalmologist to look at the back of the eye. Now, LumineticsCore from Digital Diagnostics is FDA approved to make a "Refer/No-Refer" diagnostic decision without this specialized HCP. The referral loop disappears and a patient can get their eye checked by a machine at a local grocery store clinic or a GP's office.
5️⃣ Dermatology Triage for a skin lesion usually starts with a doctor's referral to a dermatologist (often, a wait of many months). DermaSensor, Inc. - AI-Powered Skin Cancer Detection Device allows doctor to get an instant, high-accuracy assessment of whether a mole is "benign" or "malignant." Screening is put into the hands of a primary care provider or even patients via smartphone apps.
And wait for my post this Saturday on PredictiveAI, it is obliterating workflows.