Evidence-Based Practice
The clinical decision-making framework defined by ASHA as the integration of three components: best available research evidence, clinical expertise, and client/patient values and preferences. AI output is not a substitute for any of them. A model can help you find or organize information, but it cannot replace the clinician's judgment or the client's voice.
A decision-making framework originating in healthcare that requires integrating empirical evidence, practitioner expertise, and stakeholder input. In the context of AI tools, EBP serves as the standard against which AI-assisted workflows should be evaluated; the tool should support and streamline evidence-based processes, not bypass or replace them.
Why SLPs Need to Know This
The emergence of AI tools creates pressure to treat model output as a form of evidence. It is not. A model’s recommendation is a statistical pattern completion, not a clinical finding. EBP provides the framework for keeping AI in its proper role: a tool that supports your process, not a replacement for your clinical reasoning.
Clinical Impact
- Research evidence: AI can help you search for and summarize research, but it can hallucinate citations, misrepresent findings, and cannot evaluate study quality
- Clinical expertise: No model has your clinical experience, your knowledge of your setting, or your understanding of what’s feasible in your caseload
- Client values: AI has no relationship with your client and cannot incorporate their preferences, cultural context, or lived experience
- The risk: When AI output looks polished and professional, it’s tempting to skip verification. This is where EBP breaks down
Practical Guide
- Use AI to accelerate, not replace, evidence gathering. Let it help you search, but verify every source it provides
- Apply your clinical expertise to every output. If something reads well but doesn’t align with your clinical knowledge, trust your training
- Keep the client in the loop. AI-generated goals, recommendations, or materials should still reflect the client’s priorities and values
- Document your reasoning. “The AI suggested X” is not a clinical rationale. Your clinical reasoning must stand on its own.
The ASHA Triad and AI
| EBP Component | What AI Can Do | What AI Cannot Do |
|---|---|---|
| Research Evidence | Search, summarize, organize | Evaluate quality, ensure accuracy, replace peer review |
| Clinical Expertise | Surface patterns, draft documentation | Exercise judgment, understand context, take responsibility |
| Client Values | Generate culturally responsive templates | Know your client, build rapport, honor preferences |
Related Terms
- Hallucination: the primary threat to AI’s role in supporting evidence-based work
- Grounding: the technical approach that best aligns AI output with evidence-based principles
- RAG: a method for connecting AI output to specific evidence sources