— name: ai-analyzer description: AI-driven comprehensive health-analysis system. Integrates multi-dimensional health data, detects abnormal patterns, predicts health risk, and gives personalized advice. Supports intelligent Q&A and AI health-report generation. allowed-tools: Read, Grep, Glob, Write risk: unknown source: community — # AI health analyzer An AI-based comprehensive health-analysis system providing intelligent health insight, risk prediction, and personalized advice. ## When to Use – The user wants AI-driven health analysis across multiple health datasets or lifestyle signals. – You need anomaly detection, risk prediction, or personalized recommendations based on health inputs. – You need generated health reports or question-answering over health metrics and trends. ## Core features ### 1. Intelligent health analysis – **Multi-dimensional data integration**: integrates 4 data-source categories — basic metrics, lifestyle, mental health, medical history – **Anomaly-pattern detection**: detects outliers and change points using CUSUM, Z-score, etc. – **Correlation analysis**: computes correlations between health metrics (Pearson, Spearman) – **Trend prediction**: trend analysis and forecasting from historical data ### 2. Health-risk prediction – **Hypertension risk**: based on the Framingham risk-score model – **Diabetes risk**: based on the ADA diabetes risk-score criteria – **Cardiovascular-disease risk**: based on the ACC/AHA ASCVD guidelines – **Nutrient-deficiency risk**: based on RDA-attainment rate and dietary-pattern analysis – **Sleep-disorder risk**: based on PSQI and sleep-pattern analysis ### 3. Personalized-advice engine – **Basic personalization**: based on static profile — age, sex, BMI, activity level, etc. – **Advice tiers**: Level 1 (general), Level 2 (referential), Level 3 (medical advice) – **Evidence basis**: grounded in medical guidelines and evidence-based medicine – **Actionability**: concrete, feasible improvement suggestions ### 4. Natural-language interaction – **Intelligent Q&A**: supports health-data queries, trend analysis, correlation queries, etc. – **Context understanding**: maintains dialogue history, supports multi-turn conversation – **Intent recognition**: identifies the user's query intent, gives precise replies ### 5. AI health-report generation – **Comprehensive report**: all-dimension health data, AI insight, risk assessment – **Quick summary**: key-metric overview, anomaly alerts, main advice – **Risk-assessment report**: disease risks, risk-factor analysis, preventive measures – **Trend-analysis report**: multi-dimensional trends, change-point detection, predictive analysis – **Interactive HTML report**: ECharts charts, Tailwind CSS styling ## Usage ### Trigger conditions Use this skill when the user mentions any of these: **General requests:** – ✅ "AI-analyze my health" – ✅ "What are my health risks?" – ✅ "Generate an AI health report" – ✅ "AI-analyze all the data" **Risk prediction:** – ✅ "Predict my hypertension risk" – ✅ "Am I at risk for diabetes?" – ✅ "Assess my cardiovascular risk" – ✅ "AI-predict my health risks" **Intelligent Q&A:** – ✅ "How's my sleep?" – ✅ "How does exercise affect my health?" – ✅ "How should I improve my health?" – ✅ "AI health-assistant Q&A" **Report generation:** – ✅ "Generate an AI health report" – ✅ "Create a comprehensive analysis report" – ✅ "AI risk-assessment report" ### Execution steps #### Step 1: Read AI config “`javascript const aiConfig = readFile('data/ai-config.json'); const aiHistory = readFile('data/ai-history.json'); “` Check whether AI features are enabled; validate the data-source config. #### Step 2: Read the user profile “`javascript const profile = readFile('data/profile.json'); “` Get basic info: age, sex, height, weight, BMI, etc. #### Step 3: Read health data Read the relevant data per the configured data sources: “`javascript // basic health metrics const indexData = readFile('data/index.json'); // lifestyle data const fitnessData = readFile('data-example/fitness-tracker.json'); const sleepData = readFile('data-example/sleep-tracker.json'); const nutritionData = readFile('data-example/nutrition-tracker.json'); // mental-health data const mentalHealthData = readFile('data-example/mental-health-tracker.json'); // medical history const medications = exists('data/medications.json') ? readFile('data/medications.json') : null; const allergies = exists('data/allergies.json') ? readFile('data/allergies.json') : null; “` #### Step 4: Data integration and preprocessing Integrate all sources; clean data, align timestamps, handle missing values. #### Step 5: Multi-dimensional analysis **Correlation analysis**: compute links like sleep↔mood, exercise↔weight, nutrition↔biochemical markers **Trend analysis**: identify trend direction with linear regression, moving averages, etc. **Anomaly detection**: detect outliers and change points with CUSUM, Z-score #### Step 6: Risk prediction Predict risk using Framingham, ADA, ACC/AHA, etc.: – Hypertension risk (10-year probability) – Diabetes risk (10-year probability) – Cardiovascular-disease risk (10-year probability) – Nutrient-deficiency risk – Sleep-disorder risk #### Step 7: Generate personalized advice Generate three tiers of advice from the analysis: – **Level 1**: general advice (based on standard guidelines) – **Level 2**: referential advice (based on personal data) – **Level 3**: medical advice (needs doctor confirmation; includes a disclaimer) #### Step 8: Generate the analysis report **Text report**: overall assessment, risk prediction, key trends, correlation findings, personalized advice **HTML report**: call `scripts/generate_ai_report.py` to generate an interactive report with ECharts charts #### Step 9: Update AI history Log the analysis result to `data/ai-history.json` ## Data sources | Source | File path | Contents | |——–|———|———| | User profile | `data/profile.json` | age, sex, height, weight, BMI | | Medical records | `data/index.json` | biochemical markers, imaging | | Fitness tracking | `data-example/fitness-tracker.json` | exercise type, duration, intensity, MET value | | Sleep tracking | `data-example/sleep-tracker.json` | sleep duration, quality, PSQI score | | Nutrition tracking | `data-example/nutrition-tracker.json` | diet log, nutrient intake, RDA-attainment rate | | Mental health | `data-example/mental-health-tracker.json` | PHQ-9, GAD-7 scores | | Medication records | `data/medications.json` | drug name, dose, usage, adherence | | Allergy history | `data/allergies.json` | allergen, severity | ## Algorithms ### Correlation analysis – **Pearson correlation**: continuous variables (e.g. sleep duration vs mood score) – **Spearman correlation**: ordinal variables (e.g. symptom severity) ### Anomaly detection – **CUSUM**: time-series change-point detection – **Z-score**: statistical outlier detection (|z| > 2) – **IQR method**: interquartile-range outlier detection ### Risk prediction – **Framingham risk score**: hypertension, cardiovascular-disease risk – **ADA risk score**: type-2 diabetes risk – **ASCVD calculator**: atherosclerotic cardiovascular-disease risk ## Safety and compliance ### Must follow – ❌ no medical diagnosis – ❌ no specific drug-dose advice – ❌ no life-or-death prognosis – ❌ not a substitute for a doctor – ✅ all analysis must be marked "for reference only" – ✅ Level 3 advice must include a disclaimer – ✅ high-risk predictions must advise consulting a doctor ### Privacy – ✅ all data stays local – ✅ no external API calls – ✅ HTML report runs standalone ## Related commands – `/ai analyze` — AI comprehensive analysis – `/ai predict [risk_type]` — health-risk prediction – `/ai chat [query]` — natural-language Q&A – `/ai report generate [type]` — generate AI health report – `/ai status` — view AI feature status ## Implementation ### Tool restrictions This skill uses only: – **Read**: read JSON data files – **Grep**: search specific patterns – **Glob**: find data files by pattern – **Write**: generate HTML reports and update history ### Performance – Incremental reads: only read files in the specified time range – Data caching: avoid re-reading the same file – Lazy computation: generate chart data on demand ## Limitations – Use this skill only when the task clearly matches the scope described above. – Do not treat the output as a substitute for environment-specific validation, testing, or expert review. – Stop and ask for clarification if required inputs, permissions, safety boundaries, or success criteria are missing.