CMSA National Partnered with League (1 CE) – From Symptom to Signal: What the Evidence Says About AI-Supported Patient Engagement in Cancer, Palliative, and Chronic Care

Register Here!

From Symptom to Signal: What the Evidence Says About AI-Supported Patient Engagement in Cancer, Palliative, and Chronic Care

FREE WEBINAR

1 hr RN, SW, CCM

Abstract:

Case managers absorb the consequences of a fragmented delivery system. The typical Medicare beneficiary consults seven physicians across four practices annually, and direct communication between hospital physicians and primary care occurs in as few as 3–20% of discharges. The burden concentrates in the populations case management serves most intensively: 3.85 million adult 30-day readmissions each year at a mean cost of $16,300; half of older adults attend the emergency department in the final month of life; and approximately half of emergency presentations by patients with cancer are classified as preventable. These pressures coincide with a nursing workforce in which almost 40% report intention to leave within five years, and with 63 million family caregivers — the majority without formal medical training — sustaining care at home. Patients, meanwhile, have adopted new tools independently: most Americans who seek health information online now regard AI-generated content as at least somewhat trustworthy.

The trial evidence is more instructive than either enthusiasm or skepticism alone. Electronic patient-reported outcome monitoring in oncology prolonged survival and reduced emergency attendances; early palliative care delivered by video proved equivalent to in-person consultation; and the large remote-monitoring trials in heart failure and COPD that showed no benefit shared a common design flaw — no clinician with authority to act was connected to the incoming signal. The behavioral-science literature accounts for much of the remaining variance, because engagement rather than technology is the principal point of failure in digital health: defaults, well-timed prompts, tailoring, and patient activation each carry measurable effects on attendance, adherence, and cost.

This session synthesizes these literatures for case management practice. It defines agentic AI — systems that plan and execute multi-step tasks under human oversight — and sets out the governance expectations of current guidance (NIST AI RMF, WHO guidance, FDA and ONC requirements), including automation bias and equity risk. A case study of an AI-supported care platform deployed across palliative and chronic disease programs demonstrates these principles in operation, from structured symptom capture through clinician escalation and overnight caregiver support. Attendees will leave with a structured set of questions with which to appraise any AI tool used in patient care, and with a clear account of the case manager’s position as the human in the loop on which the evidence base converges.

Learning Objectives:

1. Describe the evidence for digitally supported symptom monitoring, virtual palliative care, and remote patient monitoring in oncology, palliative, and chronic disease populations — including trials that showed no benefit and the design features that separate success from failure.
2. Distinguish agentic AI from chatbots and machine-learning risk models, and identify the governance safeguards (NIST AI RMF, WHO guidance, human-in-the-loop escalation) that current guidance expects of AI in patient care.
3. Apply behavioral-science principles — defaults, timing, tailoring, and patient activation — to appraise digital patient engagement tools in their own practice.

Speaker:
Dr. Matea Deliu MD, PGDip Digital Health, PhD, MRCGP, MBCS