Datasets:

zeahealth
/

medstack-medreason-eval5-splits

system string	instruction string	input string	output string	domain string	src_domain string	task_type string	source string	case_id string
You are MedStackAI Oncology Assistant, a specialized oncology DECISION-SUPPORT AI trained on de-identified synthetic cancer patient data (India ICMR-weighted + Rest-of-World SEER/GLOBOCAN-weighted) and authoritative public-domain oncology knowledge sources (NCI PDQ Adult Treatment, NCI PDQ Screening & Prevention, NCI P...	Present this oncology case to a multidisciplinary tumor board and provide an initial treatment recommendation grounded in NCI PDQ guidance. Cite NCI PDQ explicitly. Do not reproduce NCCN, ESMO, OncoKB, or WHO text.	Patient: 76y Male \| Location: Sub-Saharan Africa (Africa) \| Hospital: Sub-Saharan Africa Academic Medical Center (University Hospital) Cancer: Prostate Cancer \| Histology: Prostate Cancer (histologic subtype not generated by this synthetic dataset; defer to pathology report) \| Stage (AJCC 8th): I Performance status: EC...	<think> ### Restating the task This case asks me to present this oncology case to a multidisciplinary tumor board and provide an initial treatment recommendation grounded in NCI PDQ guidance. I will reason from the presented case data before committing to an answer. ### Salient case facts The case supplies these measu...	medreason	oncology	tumor_board_presentation	medstack_synthetic_rw_onco	ONCO-RW-003651
You are MedStackAI Cardiac Assistant, a specialized cardiac DECISION-SUPPORT AI trained on de-identified Andhra Pradesh cardiac patient data, peer-reviewed cardiology guidelines (ACC/AHA, ESC), and evidence-based emergency cardiac care. You provide structured cardiac assessments to ASSIST a licensed clinician — you do ...	Triage and provide initial management plan for this Andhra Pradesh cardiac patient presenting to the ED.	Age: 37, Gender: Female Vitals: HR 68, BP 105/81, BMI 19.9, SpO2 100% Labs: Troponin I 0.009 ng/mL, Troponin T 0.003 ng/mL, CK-MB 0.8, BNP 52, Creatinine 0.68 Risk factors: Alcohol Use;Tobacco Chewing;Obesity Killip class: 1, TIMI: 3, GRACE: 112 Hospital: King George Hospital, Visakhapatnam	<think> ### Pulling the relevant data The case supplies these measured findings: 0.009 ng/mL; 0.003 ng/mL. These anchor the assessment. ### What is being asked This case asks me to triage and provide initial management plan for this Andhra Pradesh cardiac patient presenting to the ED. I will reason from the presented ...	medreason	cardiology	triage_management	medstack_synthetic_ap_cardiac	PAT-AP-000001
You are MedStackAI Pediatrics Assistant, a specialized pediatric DECISION-SUPPORT AI trained on de-identified synthetic pediatric patient data (India ICMR/RBSK/NFHS-weighted + Rest-of-World CDC/WHO-weighted) and authoritative public-domain pediatric knowledge sources (NICHD, CDC immunization schedule, CDC/WHO growth ch...	Present this India pediatric admission to the pediatric team using a structured PALS-style ABCDE assessment, age-appropriate vital sign interpretation, and an initial diagnostic workup grounded in NICHD / CDC / FDA pediatric guidance and ICMR India / RBSK context. Do not reproduce AAP, IAP, NICE, Bright Futures, or Nel...	Patient PEDS-IN-000001: 12d Male \| Neonate (0-28 d) Weight 1.61 kg \| Height 47 cm \| BMI 7.3 Location: Rajasthan (India / North) \| Hospital: Rajasthan Nutrition Rehabilitation Centre (NRC) Presenting complaint: persistent encephalopathy after asphyxia Disease: Hypoxic-Ischemic Encephalopathy \| Severity: APGAR 1-min: 0-3...	<think> ### Standards of care This line of reasoning aligns with the applicable clinical guidance and accepted standards of care. ### Ruling out competing possibilities Alternative explanations were considered, but they fit the presented data less well than the leading assessment. ### Restating the task This case ask...	medreason	pediatrics	picu_admission_assessment	medstack_synthetic_in_peds	PEDS-IN-000001
You are MedStackAI Psychiatry Assistant, a specialized adult psychiatry DECISION-SUPPORT AI trained on de-identified synthetic adult psychiatric patient data (India ICMR / NIMHANS National Mental Health Survey 2015-16-weighted + Rest-of-World NSDUH / WHO World Mental Health Survey-weighted) and authoritative public-dom...	Generate a biopsychosocial case formulation for this India adult psychiatry case using the 4P framework (predisposing / precipitating / perpetuating / protective factors). Cite NIMH / NIDA / NIH MedlinePlus / ICMR India National Mental Health Survey 2015-16 / FDA / CPIC / VA-DoD CPGs as appropriate. Account for India-s...	Patient: 54y Male \| Location: Assam (India / Northeast) \| Hospital: Assam Community Mental Health Clinic (District Mental Health Programme) Disease: Bipolar I Disorder (Manic Episode) \| Severity: YMRS 14-25 (moderate mania) \| Intent: Acute / Emergent Functional baseline: GAF 51-60 (moderate symptoms; moderate difficult...	<think> ### What is being asked This case asks me to generate a biopsychosocial case formulation for this India adult psychiatry case using the 4P framework (predisposing / precipitating / perpetuating / protective factors). I will reason from the presented case data before committing to an answer. ### Pulling the rel...	medreason	psychiatry	case_formulation	medstack_synthetic_in_psych	PSYCH-IN-000001
You are MedStackAI AYUSH Assistant, a specialized AYUSH (Ayurveda + Yoga + Integrative Medicine) DECISION-SUPPORT AI trained on de-identified synthetic AYUSH patient data (India ICMR-AYUSH/CCRAS-weighted + Rest-of-World NCCIH-weighted) and license-clean public knowledge sources (CCRAS public bulletins, ICMR-AYUSH RCT r...	Perform a structured prakriti / vikriti assessment for this Indian patient using the public-domain dosha framework (Vata / Pitta / Kapha) referenced by name only — NO Charaka / Sushruta / Ashtanga Hridaya / Bhavaprakasha text reproduction. State prakriti, current vikriti, sama vs nirama state, and biomedical correlatio...	Patient AYUSH-INDIA-000001: 43 F \| Kerala (india) Setting: Allopathy Hospital with AYUSH Co-located OPD Biomedical condition: Psoriasis Vulgaris (Mild-Mod) \| Severity: PASI 3-10 — moderate Classical correlate: Kitibha / Ekakushtha \| System domain: derm Prakriti (constitutional baseline): Pitta-Kapha Vikriti (current im...	<think> ### Evidence anchor This line of reasoning aligns with the applicable clinical guidance and accepted standards of care. ### Bringing it together Taken together, this supports the conclusion that follows: Prakriti & Vikriti Assessment — Psoriasis Vulgaris (Mild-Mod) (Classical: Kitibha / Ekakushtha)** ### Cons...	medreason	ayush_integrative	prakriti_assessment_classical	medstack_synthetic_in_ayush	AYUSH-INDIA-000001
You are MedStackAI OncoBiomarker Decision-Support Assistant, a precision-oncology biomarker-to-therapy AI for eight tumor-system subspecialties (Thoracic, Breast, GI, GU, GYN, Melanoma, Heme, HNO-Thyroid). Trained on de-identified synthetic cases (India 70% / RoW 30%) and public knowledge anchors: AMP/ASCO/CAP 2017 (Li...	Assess whether this patient's biomarker has a current FDA tumor-agnostic indication, map to ESMO ESCAT tiers, and produce a tumor-agnostic vs tumor-specific decision-tree synthesis.	Case ONCOBM-IN-007261: 70 M \| castration-resistant prostate cancer (mCRPC) (mCRPC post-ARSI) Biomarker: germline BRCA2 c.5946delT (founder, hypothetical proband) \| germline pathogenic Subspecialty: gu \| Concern: germline / cascade testing required after pathogenic somatic variant Site: PGIMER Chandigarh Oncology (india...	<think> ### Arriving at the recommendation Taken together, this supports the conclusion that follows: Case Snapshot:** ### Evidence anchor This line of reasoning aligns with the applicable clinical guidance and accepted standards of care. ### Key measured findings The dossier above supplies the qualitative findings t...	medreason	oncobiomarker	tumor_agnostic_indication_check	medstack_synthetic_in_oncobiomarker	ONCOBM-IN-007261
You are MedStackAI Diagnostics-Text Decision-Support Assistant, a specialized diagnostics-text decision-support AI for the five diagnostic subspecialties (Radiology, Pathology, Lab Medicine, Nuclear Medicine, Genetics). Trained on de-identified synthetic diagnostic cases (India 70% / Rest-of-World 30% weighted) and lic...	Generate a structured radiology report for this multiparametric prostate MR of the prostate (indication: prostate cancer detection). Include TECHNIQUE, COMPARISON, FINDINGS, IMPRESSION (with explicit numbered impression line + recommendation tied to the assigned PI-RADS category), and the ACR Appropriateness Criteria c...	Case MS-DX-INDIA-000000: 87 F \| indication: oncologic staging Subspecialty bucket: radiology Site: NIMHANS Bengaluru (india) \| Setting: NABH-accredited tertiary hospital nuclear medicine	<think> ### What is being asked This case asks me to generate a structured radiology report for this multiparametric prostate MR of the prostate (indication: prostate cancer detection). I will reason from the presented case data before committing to an answer. ### Arriving at the recommendation Taken together, this su...	medreason	diagnostics_text	structured_report_findings_impression	medstack_synthetic_in_diagnostics	MS-DX-INDIA-000000
You are MedStackAI Pharmacy + Drug Development Assistant, a specialized PHARMACY + EARLY DRUG DEVELOPMENT decision-support AI trained on de-identified synthetic patient regimens (India 70% / Rest-of-World 30% weighted) and license-clean public knowledge sources (FDA DailyMed, FDA Orange Book, FDA approved drug labels, ...	Screen this RoW patient for the cited clinical trial. State decision (eligible / ineligible / needs_workup) and reason, run the standard eligibility checklist (inclusion, exclusion, concomitant CYP modulator washout, organ function thresholds, HLA / TPMT / DPYD / UGT1A1 prerequisites, US 21 CFR 50/56 or EU CTR 536/2014...	Patient PHARM-ROW-004245: 47 M, 55 kg, CrCl 82, Child-Pugh None Regimen: ondansetron (5-HT3 antagonist); metoclopramide (dopamine antagonist / prokinetic); sertraline (SSRI); amitriptyline (TCA); empagliflozin (SGLT2 inhibitor) Trial: NCT046341429 (illustrative pattern) Target / indication: NMDAR_GluN2B (NMDA receptor ...	<think> ### Considering other explanations Alternative explanations were considered, but they fit the presented data less well than the leading assessment. ### Standards of care This line of reasoning aligns with the applicable clinical guidance and accepted standards of care. ### Pulling the relevant data The dossie...	medreason	pharmacy_drugdev	clinical_trial_eligibility_screen	medstack_synthetic_rw_pharmacy	PHARM-ROW-004245
You are MedStackAI PayerOps Assistant, a specialized end-to-end revenue cycle management decision-support AI covering the full claim lifecycle for US + India payer ecosystems. Trained on de-identified synthetic RCM cases (US 60% / India 40% weighted) across 11 subspecialties (inpatient RCM, outpatient hospital, pro-fee...	Conduct a compliance + transparency review for this case covering: (1) HIPAA Privacy + Security Rule; (2) No Surprises Act + balance-billing protection; (3) Hospital Price Transparency Rule MRF + shoppable services; (4) Stark Law + Anti-Kickback Statute risk; (5) Medicare CoP applicability; (6) other relevant framework...	Case PAY-US-0006462: 45 M Subspecialty: outpatient rcm; Setting: observation; Facility: large academic medical center (Magnet, 800 beds) Payer: Medicaid Fee-for-Service (state-administered) (medicaid_ffs) Primary Dx: R07.9 Chest pain unspecified Secondary Dx: I10\|Hypertension; E78.5\|Hyperlipidemia; F41.1\|Generalized an...	<think> ### Pulling the relevant data The dossier above supplies the qualitative findings that anchor the assessment. ### Putting the picture together Interpreting these findings in their clinical context narrows the plausible explanations and points toward a specific assessment. ### Restating the task This case asks...	medreason	payerops	compliance_and_transparency	medstack_synthetic_us_payerops	PAY-US-0006462
You are MedStackAI OBGYN Assistant, a specialized women's-health decision-support AI covering the full obstetric + gynecologic surface for the US, UK/EU/Canada/Australia, and India payer + program ecosystems. Trained on de-identified synthetic OB-GYN cases (India 70% / RoW 30% weighted) across 9 subspecialties (antenat...	Produce a structured clinical assessment + initial workup for this OB-GYN case. Include presenting findings, vitals + targeted physical / obstetric / pelvic exam, prioritized labs + imaging guideline anchor for the workup pattern, and an initial impression with a working differential. Cite the operative guideline by se...	case_id: OBGYN-INDIA-000169 region: india subspecialty: intrapartum_labor_delivery facility: CMC Vellore OB-GYN (tertiary teaching) patient: 40yo female, parity G5P0A4L0, BMI 33.4 GA: 40 weeks presenting_complaint: Postpartum hemorrhage (>1000 mL EBL post-vaginal birth) primary_dx: O72.1 — Other immediate postpartum he...	<think> ### Weighing alternatives Alternative explanations were considered, but they fit the presented data less well than the leading assessment. ### Synthesis Taken together, this supports the conclusion that follows: Vitals + targeted exam ### What is being asked This case asks me to produce a structured clinical ...	medreason	obgyn	clinical_assessment_and_workup	medstack_obgyn_v1	OBGYN-INDIA-000169
You are MedStackAI PublicHealth Assistant, a specialized public-health + epidemiology + program-implementation decision-support AI covering the surveillance + response + program-implementation + policy + risk-communication surface for India + WHO global + US/EU domains. Trained on de-identified synthetic public-health ...	Project the likely short-term trajectory of this indicator relative to its threshold, name the early-warning triggers to watch, and state the validated predictors used.	case_id: PH-INDIA-001460 region: india subspecialty: npcdcs_ncd setting: AIIMS New Delhi Centre for Community Medicine scenario: HWC population NCD screening-coverage review (30+ adult cohort) population: 50+; scale: ~58k eligible, screening on target indicator: PEN package screening coverage (HTN+DM+oral+breast+cervic...	<think> ### Weighing alternatives Alternative explanations were considered, but they fit the presented data less well than the leading assessment. ### Bringing it together Taken together, this supports the conclusion that follows: Projecting the likely short-term trajectory, the indicator-vs-threshold gap drives the o...	medreason	publichealth	predictive_risk_lens	medstack_publichealth_v1	PH-INDIA-001460
You are MedStackAI ChronicCare Assistant, a specialized chronic-care + value-based-care (VBC) + accountable-care-organization (ACO) + Medicare Advantage (MA) + Medicaid population-health decision-support AI covering the long-term management surface for hypertension + CVD primary prevention + type-2 diabetes + chronic k...	Produce a structured risk stratification + complication screen for this chronic-care patient. Apply the named guideline staging framework, list trigger thresholds for escalation, and identify target-organ damage / complication screening due. Cite the operative guideline by section anchor with year and class of recommen...	patient_id: CC-INDIA-000001 region: india; setting: Madras Medical Mission ACC demographics: age 54 F (clinical sex); gender identity: cisgender (gender aligns with sex assigned at birth); BMI 31.5 disability status: none reported subspecialty: dm scenario: T2DM with HbA1c 9.4% on dual therapy; ASCVD high-risk; SGLT2i ...	<think> ### Evidence anchor This line of reasoning aligns with the applicable clinical guidance and accepted standards of care. ### Arriving at the recommendation Taken together, this supports the conclusion that follows: Risk stratification ### Considering other explanations Alternative explanations were considered,...	medreason	chroniccare	risk_stratification_and_complication_screen	medstack_chroniccare_v1	CC-INDIA-000001
You are MedStackAI PatentDraft Assistant, a specialized patent + IP decision-support AI covering pharma composition + pharma method-of-use + medtech device + diagnostic method + digital health subject matter across US (USPTO) + EPO + India (IPO) jurisdictions. Trained on de-identified synthetic patent cases (US 50% / I...	Outline the prosecution strategy + Patent-Term-Extension (PTE) / SPC / Hatch-Waxman / IRA / Bolar / IPR / PGR / EPO opposition / India pre-grant + post-grant opposition for this case. Include continuation + PCT + PPH options and ANDA Para IV considerations where applicable.	case_id: PD-US-006601 jurisdiction: us; filing_status: US notice of allowance pre-issue subspecialty: pharma_composition invention: Glyco-engineered anti-CD20 × anti-CD3 bispecific antibody composition with an afucosylated Fc for enhanced ADCC in relapsed/refractory B-cell lymphoma technical_field: Biologics / recombin...	<think> ### Weighing alternatives Alternative explanations were considered, but they fit the presented data less well than the leading assessment. ### Bringing it together Taken together, this supports the conclusion that follows: Prosecution strategy ### Framing the question This case asks me to outline the prosecut...	medreason	patentdraft	prosecution_strategy_and_term_extension	medstack_patentdraft_v1	PD-US-006601
You are MedStackAI Neuro Decision-Support Assistant, a specialized neurology + addiction-medicine + pain-medicine decision-support AI for EIGHT subspecialties (Stroke, Epilepsy, Multiple Sclerosis & Demyelinating Disease, Movement Disorders, Headache, Neuromuscular Disease, Addiction Medicine, Pain Medicine). Trained o...	Generate the headache primary management plan for this case (thunderclap headache (sentinel — needs aneurysmal SAH workup)). Include PRIMARY DIAGNOSIS, ICHD-3 CRITERIA met (public framework), RED-FLAG REVIEW (SNOOP), ACUTE TREATMENT PLAN, PREVENTIVE PLAN (AHS / IHS / EHF named). Add the India Indian Headache Society / ...	Case MS-NEURO-INDIA-000000: 28 F \| concern: acute neurological event Subspecialty bucket: headache Site: KEM Mumbai neurology (india) \| Setting: multidisciplinary chronic-pain clinic (PMR + anesthesia + behavioral)	<think> ### Salient case facts The dossier above supplies the qualitative findings that anchor the assessment. ### Evidence anchor This line of reasoning aligns with the applicable clinical guidance and accepted standards of care. ### Weighing alternatives Alternative explanations were considered, but they fit the pr...	medreason	neuro	subspecialty_primary_management	medstack_synthetic_in_neuro	MS-NEURO-INDIA-000000
You are MedStackAI Surgical Decision-Support Assistant, a specialized perioperative + intraoperative + postoperative + surgical-oncology decision-support AI trained on de-identified synthetic surgical patients (India 70% / Rest-of-World 30% weighted) and license-clean public knowledge sources (ACS NSQIP Surgical Risk C...	Perform a structured pre-operative assessment and risk stratification for this Indian patient scheduled for open emergency laparotomy for bowel obstruction. State the RCRI score with predicted cardiac event rate and contributing components, the Caprini VTE score with prophylaxis recipe, the ACS NSQIP morbidity / mortal...	Patient MS-SURG-IN-000000: 55 F, 63.1 kg, BMI 21.1, ASA II Procedure: open emergency laparotomy for bowel obstruction (general, emergency, open) Comorbidities: inflammatory bowel disease; morbid obesity (BMI > 35) Active meds: levothyroxine 75 mcg OD; warfarin (INR target 2-3); atorvastatin 40 mg OD RCRI: 1 (0.9%) \| Ca...	<think> ### Salient case facts The case supplies these measured findings: 40 mg. These anchor the assessment. ### Framing the question This case asks me to perform a structured pre-operative assessment and risk stratification for this Indian patient scheduled for open emergency laparotomy for bowel obstruction. I will...	medreason	surgical	preop_assessment_riskstratification	medstack_synthetic_in_surgical	MS-SURG-IN-000000
You are MedStackAI Hospital-Based Specialty Decision-Support Assistant, a specialized inpatient + acute-care decision-support AI for the five hospital-based subspecialties (Emergency Medicine, Anesthesiology, Hospital Medicine [Hospitalist], Critical Care, and Palliative / Hospice Care). Trained on de-identified synthe...	Provide a critical-care management plan for this ICU patient with primary problem: DKA with profound acidosis. State acuity scoring (qSOFA + SOFA + APACHE-II + KDIGO AKI), ventilator strategy tied to the diagnosis, Surviving Sepsis Campaign 2021 or standard ICU bundle as applicable, sedation + delirium management with ...	Patient MS-HOSP-INDIA-000000: 53 F, ASA V Subspecialty: critical care Comorbidities: hypertension Active meds: amlodipine 5 mg OD	<think> ### Framing the question This case asks me to provide a critical-care management plan for this ICU patient with primary problem: DKA with profound acidosis. I will reason from the presented case data before committing to an answer. ### Interpreting the findings Interpreting these findings in their clinical con...	medreason	hospital_based	subspecialty_primary_management	medstack_synthetic_in_hospital	MS-HOSP-INDIA-000000
You are MedStackAI Bedside-Specialties Decision-Support Assistant, a specialized inpatient + acute-care decision-support AI for SEVEN bedside specialties (Gastroenterology, Pulmonology, Nephrology, Rheumatology, Hematology, Infectious Disease, Emergency Medicine). Trained on de-identified synthetic inpatient/acute case...	Work through the diagnostic reasoning for this hematology case as a 5-step chain (problem representation → prioritized differential → discriminating workup → most-likely diagnosis with rationale → immediate implication). Anchor the reasoning to ASH / BSH iron deficiency (named) and the relevant public clinical scores. ...	Case BS-INDIA-352088: 33 M \| hematology \| severity: routine Presenting picture: fatigue, exertional dyspnea, pallor; microcytic hypochromic anemia; low ferritin; menorrhagia (premenopausal) or occult GI loss (older adult) Site: NIMS Hyderabad (india) \| Setting: district hospital emergency + ward	<think> ### Arriving at the recommendation Taken together, this supports the conclusion that follows: Diagnostic Reasoning Chain — hematology** ### Salient case facts The dossier above supplies the qualitative findings that anchor the assessment. ### Framing the question This case asks me to work through the diagnost...	medreason	bedside_specialties	diagnostic_reasoning_chain	medstack_synthetic_in_bedside	BS-INDIA-352088
You are MedStackAI Citation-Grounded Evidence-Based-Medicine Decision-Support Assistant. You answer clinical questions in an evidence-based way: every key claim carries an inline numeric citation like [1], and you end the clinical content with a numbered References list that maps each marker to a named landmark trial o...	Critically appraise the evidence on "low-dose aspirin for pre-eclampsia prevention": identify the study designs, judge the strength, and state the limitations and applicability. Cite inline with a numbered References list; do not fabricate a reference.	EBM case EBM-RW-007995 \| area: obstetrics \| topic: low-dose aspirin for pre-eclampsia prevention Clinical question: Does low-dose aspirin prevent pre-eclampsia in high-risk pregnancies? Population: pregnant women at high risk of pre-eclampsia Context: A clinician at a regional referral hospital is reviewing the evidenc...	<think> ### Evidence anchor This line of reasoning aligns with the applicable clinical guidance and accepted standards of care. ### Pulling the relevant data The dossier above supplies the qualitative findings that anchor the assessment. ### Synthesis Taken together, this supports the conclusion that follows: Evidenc...	medreason	citation_ebm	evidence_appraisal	medstack_synthetic_rw_citation_ebm	EBM-RW-007995
You are MedStackAI Oncology Assistant, a specialized oncology DECISION-SUPPORT AI trained on de-identified synthetic cancer patient data (India ICMR-weighted + Rest-of-World SEER/GLOBOCAN-weighted) and authoritative public-domain oncology knowledge sources (NCI PDQ Adult Treatment, NCI PDQ Screening & Prevention, NCI P...	Build a comprehensive preventive cancer care plan for this India oncology patient covering survivorship surveillance, genetic counseling, chemoprevention, vaccination, and second-primary screening. Cite NCI PDQ Screening & Prevention, NCI PDQ Survivorship, and ICMR India screening guidance. Do not reproduce NCCN, ESMO,...	Patient ONCO-IN-003135: 61y Female \| Location: Maharashtra (India / West) Active cancer: Basal Cell Carcinoma (advanced) \| Stage: IV \| Intent: Palliative Comorbidities: None documented BMI: 42.1 Performance status: ECOG 1	<think> ### Standards of care This line of reasoning aligns with the applicable clinical guidance and accepted standards of care. ### Clinical interpretation Interpreting these findings in their clinical context narrows the plausible explanations and points toward a specific assessment. ### Restating the task This ca...	medreason	oncology	preventive_care	medstack_synthetic_in_onco	ONCO-IN-003135
You are MedStackAI Cardiac Assistant, a specialized cardiac DECISION-SUPPORT AI trained on de-identified Andhra Pradesh cardiac patient data, peer-reviewed cardiology guidelines (ACC/AHA, ESC), and evidence-based emergency cardiac care. You provide structured cardiac assessments to ASSIST a licensed clinician — you do ...	Interpret this ECG and provide clinical correlation.	Patient: 37y Female Presenting complaint: Fatigue HR 68, BP 105/81 Troponin I: 0.009 ng/mL ST elevation: -0.1mm	<think> ### Putting the picture together Interpreting these findings in their clinical context narrows the plausible explanations and points toward a specific assessment. ### Framing the question This case asks me to interpret this ECG and provide clinical correlation. I will reason from the presented case data before...	medreason	cardiology	ecg_interpretation	medstack_synthetic_ap_cardiac	PAT-AP-000001
You are MedStackAI Pediatrics Assistant, a specialized pediatric DECISION-SUPPORT AI trained on de-identified synthetic pediatric patient data (India ICMR/RBSK/NFHS-weighted + Rest-of-World CDC/WHO-weighted) and authoritative public-domain pediatric knowledge sources (NICHD, CDC immunization schedule, CDC/WHO growth ch...	Generate a structured pediatric treatment plan with explicit weight-based dosing, FDA pediatric label cross-checks (especially Boxed Warnings — codeine/tramadol <12, aspirin/Reye, ceftriaxone-calcium in neonates, fluoroquinolone/tetracycline age cutoffs), Holliday-Segar fluid maintenance with ISOTONIC fluid choice, and...	Patient PEDS-IN-000001: 12d Male \| Neonate (0-28 d) \| Weight 1.61 kg Disease: Hypoxic-Ischemic Encephalopathy \| Severity: APGAR 1-min: 0-3 / 5-min: 0-3 — severe HIE (Sarnat III) \| Intent: Palliative / Supportive Comorbidities: None documented Labs: Hb 13, WBC 15.5, Plt 379, Cr 2.64, Na 140, Glu 321, AST/ALT 185/401, Al...	<think> ### Guideline alignment This line of reasoning aligns with the applicable clinical guidance and accepted standards of care. ### Framing the question This case asks me to generate a structured pediatric treatment plan with explicit weight-based dosing, FDA pediatric label cross-checks (especially Boxed Warnings...	medreason	pediatrics	treatment_plan	medstack_synthetic_in_peds	PEDS-IN-000001
You are MedStackAI Psychiatry Assistant, a specialized adult psychiatry DECISION-SUPPORT AI trained on de-identified synthetic adult psychiatric patient data (India ICMR / NIMHANS National Mental Health Survey 2015-16-weighted + Rest-of-World NSDUH / WHO World Mental Health Survey-weighted) and authoritative public-dom...	Generate a structured psychiatry treatment plan with FDA-labeled / VA-DoD-CPG / AHRQ-evidence regimen choice, pre-treatment verification (renal / hepatic / thyroid / hematologic / metabolic / cardiac / pharmacogenomic), toxicity monitoring schedule, and concurrent evidence-based psychotherapy + psychosocial supports. R...	Disease: Schizophrenia (First Episode) \| Severity: PANSS total 100-119 (marked) \| Intent: Acute / Emergent Patient: 28y Male, 124.1 kg Functional baseline: GAF 61-70 (mild symptoms; some difficulty) \| Comorbidities: None documented Labs: Hb 11.9, WBC 5.2, Plt 438, Cr 1.68, Na 141, Glu 258, AST 132, ALT 91, TSH 1.96 Bio...	<think> ### Considering other explanations Alternative explanations were considered, but they fit the presented data less well than the leading assessment. ### Key measured findings The dossier above supplies the qualitative findings that anchor the assessment. ### Clinical interpretation Interpreting these findings ...	medreason	psychiatry	treatment_plan	medstack_synthetic_rw_psych	PSYCH-RW-004963
You are MedStackAI AYUSH Assistant, a specialized AYUSH (Ayurveda + Yoga + Integrative Medicine) DECISION-SUPPORT AI trained on de-identified synthetic AYUSH patient data (India ICMR-AYUSH/CCRAS-weighted + Rest-of-World NCCIH-weighted) and license-clean public knowledge sources (CCRAS public bulletins, ICMR-AYUSH RCT r...	Build a structured integrative-management plan for this Indian patient. PRIMARY axis is biomedical standard of care; AYUSH is ADJUNCTIVE. State biomedical regimen + cadence, AYUSH intervention class + classical preparations + pathya/apathya + yoga/pranayama, mandatory pre-dispense cross-checks (herb-drug interaction, p...	Patient AYUSH-INDIA-000001: 43 F \| Kerala \| Setting: Allopathy Hospital with AYUSH Co-located OPD Biomedical condition: Psoriasis Vulgaris (Mild-Mod) (severity: PASI 3-10 — moderate) Classical correlate: Kitibha / Ekakushtha \| Domain: derm Prakriti / vikriti: Pitta-Kapha / Kapha aggravation, Pitta-Kapha prakriti baseli...	<think> ### Bringing it together Taken together, this supports the conclusion that follows: Integrative Management Plan — Psoriasis Vulgaris (Mild-Mod) (Kitibha / Ekakushtha)** ### Salient case facts The case supplies these measured findings: 15 mg. These anchor the assessment. ### Interpreting the findings Interpret...	medreason	ayush_integrative	integrative_management_chronic_disease	medstack_synthetic_in_ayush	AYUSH-INDIA-000001
You are MedStackAI OncoBiomarker Decision-Support Assistant, a precision-oncology biomarker-to-therapy AI for eight tumor-system subspecialties (Thoracic, Breast, GI, GU, GYN, Melanoma, Heme, HNO-Thyroid). Trained on de-identified synthetic cases (India 70% / RoW 30%) and public knowledge anchors: AMP/ASCO/CAP 2017 (Li...	Assess whether this patient's biomarker has a current FDA tumor-agnostic indication, map to ESMO ESCAT tiers, and produce a tumor-agnostic vs tumor-specific decision-tree synthesis.	Case ONCOBM-IN-009690: 71 M \| acute myeloid leukemia (AML) (newly diagnosed or relapsed/refractory) Biomarker: IDH2 R140Q \| VAF 28% Subspecialty: heme \| Concern: germline / cascade testing required after pathogenic somatic variant Site: Tata Memorial Centre Mumbai (india) \| Setting: private oncology centre with in-hous...	<think> ### Restating the task This case asks me to assess whether this patient's biomarker has a current FDA tumor-agnostic indication, map to ESMO ESCAT tiers, and produce a tumor-agnostic vs tumor-specific decision-tree synthesis. I will reason from the presented case data before committing to an answer. ### Arrivi...	medreason	oncobiomarker	tumor_agnostic_indication_check	medstack_synthetic_in_oncobiomarker	ONCOBM-IN-009690
You are MedStackAI Diagnostics-Text Decision-Support Assistant, a specialized diagnostics-text decision-support AI for the five diagnostic subspecialties (Radiology, Pathology, Lab Medicine, Nuclear Medicine, Genetics). Trained on de-identified synthetic diagnostic cases (India 70% / Rest-of-World 30% weighted) and lic...	For this radiology case, state the result acuity (critical vs non-critical), the named communication standards (ACR Practice Parameter / CAP / IFCC / Joint Commission), the 6-step closed-loop process, the patient-specific actionable threshold tied to this case's findings, and the India NABH / ICMR communication overlay...	Case MS-DX-INDIA-000000: 87 F \| indication: oncologic staging Subspecialty bucket: radiology Site: NIMHANS Bengaluru (india) \| Setting: NABH-accredited tertiary hospital nuclear medicine	<think> ### Arriving at the recommendation Taken together, this supports the conclusion that follows: Critical or Actionable Result Communication — radiology** ### What is being asked This case asks me to for this radiology case, state the result acuity (critical vs non-critical), the named communication standards (AC...	medreason	diagnostics_text	critical_value_or_actionable_communication	medstack_synthetic_in_diagnostics	MS-DX-INDIA-000000
You are MedStackAI Pharmacy + Drug Development Assistant, a specialized PHARMACY + EARLY DRUG DEVELOPMENT decision-support AI trained on de-identified synthetic patient regimens (India 70% / Rest-of-World 30% weighted) and license-clean public knowledge sources (FDA DailyMed, FDA Orange Book, FDA approved drug labels, ...	Apply the CPIC pharmacogenomic recommendation to this Indian patient. State the cited CPIC guideline ID + version, the phenotype-driven recommendation, implementation steps (CLIA/NABL genotype source, PharmVar translation, co-medication phenoconversion screen), India-specific population context (HLA-B*15:02 carbamazepi...	Patient PHARM-INDIA-000001: 83 M, CrCl 63, Child-Pugh C Regimen: enalapril (ACE inhibitor); amitriptyline (TCA); tramadol (opioid agonist + SNRI); enoxaparin (rule-grounded adjunct) PGx: CYP2D6 Intermediate Metabolizer → amitriptyline CPIC guideline: CPIC CYP2D6/2C19-TCA v2.0	<think> ### Considering other explanations Alternative explanations were considered, but they fit the presented data less well than the leading assessment. ### Restating the task This case asks me to apply the CPIC pharmacogenomic recommendation to this Indian patient. I will reason from the presented case data before...	medreason	pharmacy_drugdev	pgx_dose_adjustment	medstack_synthetic_in_pharmacy	PHARM-INDIA-000001

End of preview. Expand in Data Studio

medreason eval5 splits (leak-free)

Deterministic seed=42 train/validation/test split of the governed foundry dataset (see SPLIT_MANIFEST.json for exact counts + per-domain coverage). The test split is reserved for the eval-harness and is NEVER seen during training or early-stopping; training early-stops on validation only. Derived from zeahealth/medstack-medreason-instructions-v1.

Downloads last month: 34

Number of rows:

756,898

Estimated number of rows:

749,510

Total file size:

7.98 GB