From the Guidelines
A guideline-directed medical AI app should be developed and implemented to provide healthcare professionals with real-time decision support, integrating evidence-based clinical guidelines and condition-specific algorithms to improve patient care and outcomes. When developing such an app, it is crucial to consider the quality of guidelines and their potential impact on patient care, as highlighted in a recent systematic review 1. The app should feature the following key components:
- Condition-specific algorithms that recommend appropriate medications, dosages, and treatment durations based on patient-specific factors
- Comprehensive medication databases with dosing adjustments for renal/hepatic impairment, drug interaction alerts, and contraindication warnings
- Integration with electronic health records to allow automatic patient data import and documentation of AI-suggested interventions
- Continuous updates to the underlying AI, using high-quality guideline data from major medical societies, to ensure that the app remains current and effective The use of guideline-directed medical AI apps can improve clinical care by reducing practice variation, enhancing guideline adherence, and providing evidence-based recommendations at the point of care, while still allowing clinician judgment to override suggestions when appropriate for individual patient circumstances, as emphasized in the systematic review 1. Key considerations for the development and implementation of such apps include:
- Ensuring the quality and accuracy of the guidelines and algorithms used
- Providing transparent and explainable AI decision-making processes
- Addressing potential biases and limitations in the data used to train the AI
- Continuously monitoring and evaluating the app's performance and impact on patient outcomes, as recommended by the AGREE II and RIGHT tools 1.
From the Research
Guideline Directed Medical AI App
- The development of guidelines and standard frameworks for artificial intelligence (AI) in medicine is crucial for its integration into clinical settings 2.
- The FUTURE-AI framework provides guidance for the development and deployment of trustworthy AI tools in healthcare, based on six guiding principles: fairness, universality, traceability, usability, robustness, and explainability 3.
- For the management of hypertension, lifestyle modification is the first-line therapy, including weight loss, healthy dietary pattern, physical activity, and moderation or elimination of alcohol consumption 4, 5.
- First-line drug therapy for hypertension consists of a thiazide or thiazide-like diuretic, an angiotensin-converting enzyme inhibitor or angiotensin receptor blocker, and a calcium channel blocker, which should be titrated according to office and home blood pressure levels to achieve target blood pressure levels 4, 6.
- The use of AI in healthcare requires consideration of ethical, environmental, and reproducibility aspects, and the development of integrated and comprehensive reporting guidelines is essential for fostering a cultural shift toward transparency and open science 2.
Key Considerations for AI App Development
- The Appraisal of Guidelines, Research, and Evaluation II tool can be used to assess the quality of guidelines for AI in medicine, based on domains such as scope and purpose, stakeholder involvement, rigor of development, clarity of presentation, applicability, and editorial independence 2.
- Well-established initiatives such as TRIPOD+AI, DECIDE-AI, SPIRIT-AI, and CONSORT-AI have shown high quality in terms of stakeholder involvement, but applicability remains a prominent challenge among guidelines 2.
- The FUTURE-AI guideline provides a set of 30 best practices for the development and deployment of trustworthy AI tools in healthcare, addressing technical, clinical, socioethical, and legal dimensions 3.