Chakudya MCP Server

Categorize a GLIM-diagnosed malnutrition into one of four etiology-based categories per Table 5, to help guide intervention and anticipated outcomes: chronic disease with inflammation, chronic disease with minimal/no perceived inflammation, acute disease or injury with severe inflammation, or starvation-related (hunger/food shortage from socioeconomic or environmental factors). These categories are not mutually exclusive in real patients; provide your best clinical read of the dominant driver.

Determine whether an adult patient already flagged 'at risk' by a validated screening tool (MUST/MNA-SF/NRS-2002/etc) meets GLIM diagnostic criteria for malnutrition, per Table 3. Diagnosis requires at least one phenotypic criterion (weight loss, low BMI, reduced muscle mass) AND at least one etiologic criterion (reduced food intake/assimilation, disease burden/inflammation). Weight loss and BMI are auto-classified from numbers; reduced muscle mass (assessed by DXA/BIA/CT/MRI or, where unavailable, physical exam or MUAC/calf circumference) is taken as an already-assessed clinical judgment.

Grade the severity of a GLIM-diagnosed malnutrition (see glim_malnutrition_diagnosis) as Stage 1 (moderate) or Stage 2 (severe), per Table 4. Only phenotypic criteria (weight loss, BMI, reduced muscle mass) are used for severity — etiologic criteria inform intervention, not severity. Only one phenotypic criterion needs to reach a given stage's threshold for the patient to be graded at that stage; severe outranks moderate.

Compute Estimated Energy Requirement using the Institute of Medicine (2002/2005) DRI prediction equations (Nelms/Ireton-Jones Box 2.1), covering life stages beyond simple Mifflin-St Jeor: infants 0-2y (weight only), normal-weight children 3-18y (5th-85th BMI percentile), overweight children 3-18y (>85th BMI percentile), adults 19+ at three equation variants (normal-weight-only BMI 18.5-25, overweight/obese-only BMI>=25, and a combined normal+overweight/obese BMI>=18.5 equation set), and pregnant/lactating women (14-18y or 19-50y, which layer on top of the corresponding adolescent/adult female …

Look up the DRI acceptable macronutrient distribution range (% of kcal from carbohydrate, fat, protein) for a pediatric age group (BND 415 Paediatric Medicine Resources), and optionally check whether a proposed diet's percentages fall inside those ranges.

Get general educational information about a medication's nutrition-relevant profile — e.g. known food-drug interactions, nutrient depletion effects, general dietary considerations — grounded on the CNR RAG knowledge base with citations. This tool NEVER returns dosing, administration instructions, or prescribing guidance; it is strictly educational background, not clinical decision support for prescribing.

Estimate kcal expended during a physical activity using its metabolic equivalent (MET) value: kcal = MET x weight(kg) x duration(hours). Pick a known activity from Table 2.4 (Nelms/Ireton-Jones) via `activity`, or supply a custom `met_value` for an activity not in the table.

Classify malnutrition status for a non-pregnant, non-postpartum adult (18+) using NACS cutoffs: bilateral pitting edema, MUAC, BMI, and/or confirmed unintentional weight loss >10% since last visit (a severe-malnutrition trigger). Note the guide states WHO has not established standard MUAC cutoffs for adults — the MUAC cutoffs used here are suggestions based on current practice, not a WHO standard, unlike BMI.

Classify acute malnutrition status for a child 0-59 months old using NACS cutoffs: bilateral pitting edema, MUAC (6-59 months only), and/or weight-for-height z-score (WHZ, 6-59 months). Supply whichever measures were taken; at least one of muac_mm or whz is required unless edema is present. Returns a classification per indicator plus an overall status (edema, if present, always makes the overall classification 'severe' — this is SAM regardless of MUAC/WHZ).

Classify malnutrition status for a child/adolescent aged 5-17 years using NACS cutoffs: bilateral pitting edema, age-banded MUAC (5-9y, 10-14y, 15-17y), and/or BMI-for-age z-score. age_years is required if muac_mm is supplied, to select the correct MUAC band.

Classify malnutrition status for a pregnant or postpartum woman using NACS cutoffs: bilateral pitting edema, MUAC, and/or confirmed unintentional weight loss >10% since last visit (a severe-malnutrition trigger). The MAM MUAC upper cutoff varies by country protocol (220 or 230mm) — pass moderate_muac_upper_mm if your country uses 230mm; defaults to 220mm.

Compute BMI, Basal Metabolic Rate (Mifflin-St Jeor equation), and Total Daily Energy Expenditure from weight, height, age, sex, and activity level. Pure calculation — does not call the CNR API. Educational/clinical-support estimate only, not a substitute for individualized dietetic assessment.

Search packaged/branded food products. Pass a barcode for an exact lookup, and/or a free-text query to search by product name (substring match). At least one of barcode or query must be provided.

Estimate pediatric BMR and/or Total Energy Expenditure using one of four published methods: 'schofield_bmr' (Schofield 1985, needs weight_kg + height_cm), 'who_bmr' (WHO 1985, needs weight_kg only), 'dri_fao_2004' (direct kcal/kg/d TEE table lookup by age, needs weight_kg), or 'dri_iom_2006' (DRI/IOM 2006 EER equations — infants 0-3y need weight_kg only; children 3-18y also need height_cm and physical_activity_level). For schofield_bmr/who_bmr, optionally supply activity_factor and/or stress_factor (from the published ranges: activity 1.0-2.15, stress 0.7-2.5) to scale BMR up to an estimated T…

Look up initiation, advancement, and goal volume guidance for pediatric enteral feeds (continuous or bolus) by age, from the Pediatric and Nutrition Support Handbook, 3rd edition (2024).

Compute daily maintenance fluid requirement for a child using the Holliday-Segar method, given body weight in kg. Returns total mL/day and mL/hr.

Look up expected/reference growth velocity (weight gain, linear growth) from the ASPEN Paediatric and Neonatal Nutrition Support Handbook, 3rd edition. Set preterm=true for the preterm infant reference (initial weight loss, weight/length/HC velocity — no age input needed). Otherwise pass age_months (0-24, sex-specific) or age_years (2-11, not sex-specific) for term infants/children.

Look up pediatric protein requirement (g/kg/d) from a published source and compute total g/day for a given weight. Sources: 'iom_2005' (healthy growth RDA, by age in months/years), 'aspen_sick' (ASPEN dosing guideline for sick children, by age in years), or 'preterm' (ASPEN enteral preterm infant protein by birth/current weight in grams — pass weight_kg as normal, it is converted).

Look up estimated enteral fluid and energy requirements for preterm infants (BND 415 Paediatric Medicine Resources). Returns the reference mL/kg/d and kcal/kg/day ranges, plus totals for the given weight using the range midpoints. Pair with pediatric_protein_requirements(source='preterm') for the matching protein target.

Semantic search over the Chakudya RAG knowledge base (includes ingested nutrition textbook content) and return the raw matching chunks — no LLM-generated answer, just the ranked source passages. Use this when you want to read/cite the underlying material yourself rather than get a synthesized answer.

Look up foods in the renal diet exchange list (based on the RenalSmart South African exchange system) — potassium, phosphorus, and sodium content relevant to chronic kidney disease / dialysis diets.

Interpret a measured respiratory quotient (RQ = VCO2/VO2) from indirect calorimetry against the reference fuel-mixture values: 1.0=carbohydrate, 0.85=mixed diet, 0.82=protein, 0.7=fat, <=0.65=ketone production. Values >1 suggest net fat synthesis / excess carbohydrate or total caloric intake; very low values may indicate inadequate nutrient intake. Note: per McClave et al (2003), RQ has low sensitivity/specificity for detecting over/underfeeding in practice and is best used as a marker of measurement validity (confirming RQ is in the physiologic 0.65-1.2 range).

Retrieve a broader evidence base for a nutrition topic — runs the same grounded RAG orchestrator as search_guidelines but with context='both' (clinical + general sources) and a higher default top_k, for when you need wider supporting evidence rather than a narrow clinical-guideline answer.

Search the Chakudya Nutrition Registry (Malawi food composition database) by name. Searches the local database first; if nothing matches locally, falls back to the external lookup cascade (USDA FoodData Central / Open Food Facts / FatSecret). Use this to find a food before calling get_food_details or calculate_nutrients.

Ask a clinical nutrition guideline question and get a grounded answer with numbered citations, generated by the Chakudya RAG orchestrator (semantic search + rerank + LLM answer). Best for questions like 'what is the recommended protein intake for a stage 3 CKD patient' where you want a synthesized, cited answer rather than raw passages.

Look up the SGA body-composition guidance descriptors (Normal / Mild-Moderate / Severe) for a given physical exam site — subcutaneous fat, muscle wasting, or fluid retention — to standardize exam documentation language. Pure reference lookup; does not classify a patient.

Structure a Subjective Global Assessment by classifying each domain (nutrient intake, weight change, symptoms, functional capacity, physical exam of fat/muscle/fluid, metabolic requirement) against the SGA form's descriptive A/B/C criteria, and surface a suggested overall rating (worst-domain-dominant) as a starting point. SGA is explicitly a gestalt clinical judgment, not an additive score — the final rating, and any trajectory-based override (recent clear improvement or recent significant deterioration), is the clinician's call, not this tool's.

Estimate Total Energy Expenditure from a known/measured REE using the simplified activity-level banding method (Nelms/Ireton-Jones Ch.2): minimal activity = REE x 1.10-1.20, moderate = REE x 1.25-1.40, strenuous = REE x 1.45-1.60. These ranges are described in the source as expert opinion rather than evidence-based, and are a quicker bedside alternative to the full IOM/DRI prediction equations when you already have a predicted or measured REE.

Description is built at runtime.