Past Medical History
No known chronic medical conditions. No previous hospitalizations or surgeries. Up-to-date on vaccinations.
AI clinical reasoning simulator
Clinical thinking, Simulated patients
Prepare for OSCEs and clinical practice.
Exam-pack readiness reports for students. Cohort pilots for faculties.
Clinical education partners
Simulation, medical education and teaching-hospital partners used in how students practise the clinical encounter.
_logo.svg){kind=logo}
For students
Handle the patient before exam day.
Each case carries a scoring file written before you open it: essential questions, essential examinations, essential tests, the accepted diagnosis and its aliases, and a management plan. Your score is computed against that file. The language model writes the narrative feedback; it does not decide the marks.
Weighted score
Diagnosis
30%
Working diagnosis against the accepted answer and its aliases
Management
25%
Immediate, definitive and follow-up steps; contraindications
Investigations
20%
Tests and examinations ordered, and how you interpreted them
History
15%
Coverage of the essential questions for this presentation
Communication
10%
Empathy, clarity, patient-centredness, urgency explained
Assessed outside the score:Urgency recognitionSafety flagsDistractor handling in complex cases
Exam pack
OSCE-style stations for the week before an exam. Each student works their own cases, and every case is marked on a fixed rubric.
One readiness score
A single score across diagnosis, management, investigations, history and communication, so students know where they stand.
Shareable PDF
A one-page readiness report students can keep and send to each other. Faculties can ask for the same page across a whole year.
For universities
What each student is still missing.
A faculty can see which part of the encounter each student misses most, and teach to that weakness.
- Where it fits
- Formative practice alongside clinical rotations, pre-session preparation for flipped teaching, or a standardised exercise set for a whole year group. Cases are grouped by specialty and can be assigned by presentation.
- What faculty get
- A lecturer mode that unlocks complex cases and differential-diagnosis views. Cohort-level reporting is being built with our pilot faculties and shaped by what they ask for.
- How content is governed
- Cases are structured from clinical education, simulation and research materials, then reviewed before activation. Each case carries its own rubric. Errors reported by faculty are corrected at source, so every student sees the fix.
- Data and compliance
- Student accounts hold name, email and institution. Practice transcripts are stored to generate reports and progress. GDPR-aligned processing; a data processing agreement and hosting details are available on request.
- Pilot programme
- A pilot runs for one module with one cohort. We agree the case list and the outcome measure with you first, and we share the full data at the end, whatever it shows.
Lab
One case, six decisions, one report.
Every case follows the same structure a real encounter does. You decide what to ask, what to examine, what to order and what to do. The rubric is fixed before you start.
Assessed under
Meet the patient
Chat qualityA presenting complaint and a patient who answers in their own words. Ask what you would ask on the ward; nothing is pre-filled.
Examine
Analysis coverageRequest specific physical examination findings. Irrelevant examinations return normal results and cost you time, not marks.
Order investigations
Analysis coverageLaboratory panels return values with reference ranges. Instrumental studies return real imaging from the NLM Open-i archive.
Commit to a diagnosis
DiagnosisName your working diagnosis. Complex cases include distractor diagnoses that are plausible from the history alone.
Decide management
TreatmentImmediate, definitive and follow-up steps, with contraindications checked against the patient in front of you.
Read your report
All domainsA weighted score, a breakdown by domain, the essential questions and tests you missed, and where you were unsafe.
Case study
How one station prepares a student.
The simulation is the encounter, not a summary of it. A student has to ask, examine, investigate and decide. The report shows what they covered and where they were unsafe, so they can do it again on a different case.
Chest pain · cardiology
Before the OSCE
Ask without a script
The patient gives a presenting complaint in their own words. The student has to find the history that matters. Nothing is pre-filled.
Examine and order only what they need
They request specific findings and tests. An irrelevant examination returns a normal result. Laboratory values come back with reference ranges.
Name the diagnosis and the plan
They commit to a working diagnosis and to what they would do now, next and in follow-up. On harder cases, a plausible distractor is there if they stop at the history.
See the miss
The report marks what they asked, what they ordered, the diagnosis and the plan, and it flags an unsafe step even when the rest of the station went well.
This helps students to
- Take a history the way an OSCE requires it, with no checklist on the screen.
- Choose examinations and tests for the patient in front of them.
- Commit to a diagnosis and a management plan before the answer is given.
- Find the gap — a missed question, a missed test, an unsafe step — and practise it on another case.
Cases are unlimited, generated across disease categories and the clinical situations a student will meet.
Try a cardiology caseSources and governance
Reference-backed cases, fixed rubrics and visible provenance.
Content is structured from reference sources
Each case is structured from medical reference content into a fixed case file: presentation, findings, investigation results, accepted diagnosis and management. Imaging is drawn from the NLM Open-i open-access archive.
The patient cannot invent facts
During the encounter the patient model answers from the case file. It does not browse the web and it does not improvise findings that are not in the file.
Scores are deterministic
Marks are computed by a rubric scorer (version grounded-v1) against the case file: essential questions, examinations, tests, the diagnosis and its aliases, the management plan. The language model writes the explanation of the score. It does not set it.
Safety is assessed separately
Urgency recognition and safety flags are recorded outside the weighted score so that a dangerous decision is visible even on an otherwise good attempt.