Tier 3

aba - AI Biomedical Agent

AI Biomedical Agent

Input: $ARGUMENTS

Overview

Systematic procedure for analyzing biomedical questions using structured evidence evaluation. Searches literature, assesses study quality, synthesizes findings, and identifies limitations. Does NOT provide medical advice — produces structured analysis for informed decision-making.

Step 0: Scope and Safety Check

  1. State the biomedical question precisely
  2. Classify: basic science / clinical / pharmacological / epidemiological
  3. Safety gate: If the question is “should I take/stop medication X?” — output is ANALYSIS ONLY, not recommendation. Flag explicitly.
  4. Identify what kind of evidence would answer this (RCTs, meta-analyses, case studies, mechanistic)

Steps

Step 1: Define the Research Question

  1. Convert to PICO format where applicable:
    • Population: Who/what is being studied?
    • Intervention: What treatment/exposure?
    • Comparison: Compared to what?
    • Outcome: What result matters?
  2. If not clinical, define: Subject, Mechanism, Evidence type needed
  3. Identify key terms and synonyms for search

Step 2: Search and Gather Evidence

  1. Identify source hierarchy:
    • Systematic reviews / meta-analyses (highest)
    • Randomized controlled trials
    • Observational studies (cohort, case-control)
    • Case reports / expert opinion (lowest)
  2. Search: key terms, MeSH terms, known databases
  3. Note what you can and cannot access
  4. Flag if evidence base is thin (<3 relevant studies)

Step 3: Assess Evidence Quality

For each piece of evidence:

  1. Study design strength
  2. Sample size and statistical power
  3. Bias risk: selection, performance, detection, attrition, reporting
  4. Conflict of interest (funding, author affiliations)
  5. Replication status
  6. Assign tier: A (strong) / B (moderate) / C (weak) / D (very weak)

Step 4: Synthesize Findings

  1. What do highest-quality studies show?
  2. Consensus or disagreement? If disagreement, what explains it?
  3. Effect size (not just significance — how large?)
  4. Confidence intervals
  5. Dose-response relationship if applicable

Step 5: Assess Drug Interactions (if applicable)

  1. Identify all substances mentioned
  2. Check interaction mechanisms: CYP450, protein binding, pharmacodynamic
  3. Severity: major / moderate / minor / theoretical
  4. Evidence basis for each interaction

Step 6: Identify Limitations and Gaps

  1. What questions remain unanswered?
  2. What populations are understudied?
  3. Where is evidence weakest?
  4. What would change the conclusion if discovered?

Step 7: Report

BIOMEDICAL ANALYSIS
Question: [PICO or structured question]
Evidence tier: [A/B/C/D overall]

Key findings:
1. [finding] — Evidence: [tier] — Source: [type]
2. [finding] — Evidence: [tier] — Source: [type]

Synthesis: [what the evidence overall suggests]
Confidence: [high/moderate/low] — [reasoning]

Limitations:
- [limitation 1]
- [limitation 2]

NOT medical advice. Analysis only.

When to Use

  • Evaluating biomedical research claims
  • Understanding drug mechanisms or interactions
  • Interpreting clinical study results
  • Assessing evidence for health interventions

Verification

  • Question stated in structured format
  • Evidence sources identified with quality tiers
  • Study quality assessed (not just cited)
  • Effect sizes reported (not just significance)
  • Limitations explicitly stated
  • No medical advice given — analysis only