The next phase of healthcare AI is moving into the core of care pathways, spanning before diagnosis, across fragmented records, during encounters, inside treatment decisions, between appointments, through administrative workflows, and across populations, with the aim of enabling earlier intervention, reducing friction, improving clinician access to relevant data, and better identifying appropriate patient populations, including rare diseases.

Adoption and governance challenges accompany this shift, as NHS clinicians use AI alongside or outside formal guidelines, concerns about AI-generated errors in records grow, regulators seek formal oversight for AI in medical devices, and public demand for transparency rises.

The digest reflects September 2026 AI and healthcare research findings and company performance as described in source materials.

Innovations in treatment and care include AI models predicting post-heart-attack recovery, forecasting dangerous post-AMI bleeding with an explainable six-point score, digital twins for lung cancer predicting immunotherapy response, and AI systems that uncover untreated patients by mining unstructured notes.

Generative AI is expanding beyond drafting to streamline workflows, including referral management that automates a large share of referrals, decision support, patient-facing and clinician-facing AI in portals and dictation, and supervisory AI for governance.

By 2026, AI in healthcare is widely deployed across the patient journey, transitioning from potential to tangible outcomes, with governance and rapid adoption as central questions for health systems.

Prediction and diagnosis advances see AI flag potential issues before formal diagnosis by analyzing routine data, including ECG-based triage for heart failure and valve disease, mammograms revealing cardiovascular risk, sleep data linked to mortality risk, urine tests for pancreatic cancer, and accelerated bladder cancer detection from GP records with high accuracy.

Fracture and rare disease detection are enhanced by AI, with FRACTURE-ML identifying hip fracture risk more effectively than traditional methods and AI helping spot rare or ultra-rare disease patients earlier by analyzing real-world data and patient journeys.