Contents · 9 sections
- What is a student case history and why is it written
- Case history template: sections according to the classical scheme
- How to write a case history using an AI: step-by-step
- Student case history: example by sections
- Cross-checking with clinical guidelines and ICD-10
- What to do yourself and what can be entrusted to AI
- Medical confidentiality and personal data: what must not be uploaded to AI
- How not to get caught submitting someone else's work
- FAQ
You can write a case history using AI if the AI acts as an assistant: it organizes your patient observation notes according to the department's template, checks the diagnosis and treatment against clinical guidelines, and suggests what was missed. You must not submit the text without checking: during the defense, the instructor will ask why the diagnosis is set this way, and you will have to answer. Below is the student case history template, a section-by-section example, the workflow with AI, and the rules on how not to violate medical confidentiality.
What is a student case history and why is it written
A student (academic) case history is not a medical document like form 003/y. It is an educational assignment where a student demonstrates the ability to interview a patient, perform a physical examination, identify syndromes, substantiate a diagnosis, and formulate a treatment plan. It is written in propaedeutics of internal diseases, faculty and hospital therapy, surgery, pediatrics, neurology, infectious diseases, and obstetrics and gynecology.
Each department has its own template: some require Latin prescriptions, others require observation notes for five days of care, and some require a detailed differential diagnosis. However, the core skeleton is almost identical everywhere, and this is precisely what is convenient to assemble using an AI.
Case history template: sections according to the classical scheme
Below is the classical sequence of sections. Check it against your department's methodological guide: the order and titles may vary.
| Section | What to write | Common mistake |
|---|---|---|
| General patient data | Age, sex, occupation, date of admission and observation, referral diagnosis. In student work, data is modeled or anonymized | Real full name, address, and insurance policy number copied |
| Complaints | Chief and general complaints, upon admission and at the time of observation, with details: character, localization, radiation, provoking and relieving factors | "Chest pain" without characteristics |
| History of present illness | Chronology from the first symptoms: when it started, what examinations and treatments were performed with what effect, why hospitalized now | No dates and sequence |
| Past medical history | Previous illnesses and surgeries, bad habits (smoking index), occupational hazards, allergic, hereditary, and epidemiological history | Allergy history forgotten |
| Objective status | General inspection, then by systems: respiratory, cardiovascular, digestive, urinary, endocrine, nervous. For surgery, local status separately | "No pathology" in a single line for all systems |
| Preliminary diagnosis and examination plan | Leading syndromes, preliminary diagnosis, laboratory and instrumental methods with the purpose of each | Examination plan "just in case", without purpose |
| Examination results | Laboratory and instrumental data with interpretation | Numbers without conclusion |
| Differential diagnosis | 2–4 diseases with a similar clinical picture and the signs by which you rule them out | List of diseases without arguments |
| Clinical diagnosis and substantiation | Primary disease, complications, comorbidities, ICD-10 code. Substantiation: which complaints, history, examination, and tests confirm the diagnosis | Formulation not according to classification |
| Treatment | Regime, diet, medications with doses and justification for the choice, prescriptions if required by the department | Treatment scheme from an outdated textbook |
| Observation notes | Date, complaints, condition, respiratory rate, heart rate, blood pressure, temperature, dynamics, changes in prescriptions | Identical notes for all days |
| Prognosis and epicrisis | Observation summary, prognosis, discharge recommendations | Epicrisis repeats the history |
How the template differs across propaedeutics, therapy, surgery, and pediatrics
In propaedeutics, the main focus is on examination techniques and syndromes; the treatment section may be abbreviated or omitted. In internal medicine, the emphasis is on differential diagnosis and treatment justification based on clinical guidelines. In surgery, local status, indications for surgery, and the surgical protocol are added. In pediatrics, the mother's pregnancy and childbirth history, feeding, vaccinations, and physical and neurodevelopmental assessment are included. Therefore, always start by taking your department's specific template.
How to write a case history using an AI: step-by-step
- Take the department template. The platform has department templates added by peers. If yours is not there, work with the classical template or attach your own.
- Gather your observation notes. Text, a file, a photo of handwritten notes, or an audio recording where you dictated the interview and examination data will work.
- Anonymize the data. Remove full names, dates of birth, addresses, policy numbers, and anything that could identify the patient. More on this below.
- Run the generation. The AI organizes the data into sections of the template, formulates the diagnosis, cross-checks the diagnosis and treatment against the clinical guidelines of the Russian Ministry of Health, and inserts the ICD-10 code.
- Check each section. Everything written in the objective status must match what you actually saw and heard in the patient.
- Export as Word or PDF and format it according to the department's requirements: title page, font, reference list.
Option 1: based on case-curation notes
The main and most honest path. You curated a patient, recorded complaints, anamnesis, and examination data, and the neural network helps turn disjointed notes into a coherent text following a template. Such a case history is easy to defend: the patient is yours, the findings are yours.
Option 2: disease-based only
If you need to understand what a case history looks like for a specific nosology, you can specify only the disease. In this case, the patient is modeled: this is not a real person, but a typical clinical presentation. This option is suitable for practice, for understanding the structure and logic of justification. If the department requires the curation of a real patient, a model case history will not replace it.
In WhiteCoat Pro, 2 case histories per month are available for free on any account, and another 5 are added to them on the "Maximum" tariff. This feature is included in the neural network for medical professionals along with study notes, cheat sheets, presentations, and answers based on clinical guidelines. You can also request a case history directly from the Claude or ChatGPT chat if you connect the school to them.
Student case history: section-by-section example
The fragments below are model ones: the patient is fictionalized for illustration. Topic: community-acquired pneumonia.
General information. Patient M., 54 years old, engineer.
Complaints of cough with mucopurulent sputum, fever up to 38.7 °C, shortness of breath on exertion, pain in the right hemithorax on deep inspiration, general weakness.
History of present illness. Fell ill acutely 4 days ago after hypothermia: chills, fever; a cough appeared a day later. Took antipyretics without effect, went to the outpatient clinic, and was referred for hospitalization.
Objective status (respiratory system). RR 22 breaths per minute, SpO2 94%. The right hemithorax lags behind in the act of breathing. On the right below the scapular angle: enhanced vocal fremitus, dull percussion sound, bronchial breathing, resonant fine moist rales.
Clinical diagnosis. Community-acquired pneumonia of the lower lobe of the right lung, non-severe course. Respiratory failure I. ICD-10 code: J18.9.
Justification. The diagnosis is made on the basis of an acute onset with fever, cough with sputum and pleuritic pain, the lung tissue consolidation syndrome upon examination, and infiltration in the lower lobe on the right on the radiograph. CRB-65 score is 0.
Pay attention to the logic: each element of the diagnosis is confirmed by specific data from the previous sections. The ICD-10 code is refined if the pathogen is verified: for example, pneumonia caused by Streptococcus pneumoniae is coded as J13. Severity scales, such as CURB-65, are convenient to calculate in the clinical scales section.
Cross-checking with clinical guidelines and ICD-10
Instructors are increasingly checking whether the diagnosis and treatment comply with the current clinical guidelines of the Russian Ministry of Health. What to cross-check:
- Diagnosis formulation according to the "Classification" section of the required clinical guidelines.
- Examination plan according to the "Diagnostics" section: which examinations are recommended and with what level of evidence.
- Treatment according to the "Treatment" section: drug groups, doses, duration.
- ICD-10 code: codes are indicated on the title page of the clinical guidelines.
- Age category: for a single disease, there may be separate guidelines for adults and children.
The WhiteCoat Pro neural network performs this check when compiling the case history, but the final decision is yours. Open the full text of the required guidelines in the clinical guidelines section: it contains all 756 active clinical guidelines of the Russian Ministry of Health, search by title and ICD-10 code, text open without registration. We discussed how to quickly find an answer in clinical guidelines in the article neural network for clinical guidelines.
What to do yourself and what you can delegate to the neural network
| You can delegate to the neural network | Things you do |
|---|---|
| Organize notes into template sections | Curate the patient: questioning, examination, palpation, percussion, auscultation |
| Turn notes into coherent text with correct terminology | Verify that each finding actually occurred |
| Suggest a range of differential diagnosis | Choose what to rule out and explain why |
| Cross-check diagnosis and treatment with clinical guidelines, assign the ICD-10 code | Recheck doses and contraindications for your patient |
| Format the document in Word or PDF | Prepare for the defense |
Typical neural network errors in a case history
- "Fills in" normal data for systems you did not examine. If there is no neurological status in your notes, it should not appear on its own.
- Contradictions between sections: no shortness of breath mentioned in complaints, but it "decreased" in the follow-up notes.
- Outdated treatment regimens in universal chatbots that answer from the model's memory. Why this happens is discussed in the article about ChatGPT for medical professionals.
- General phrases that apply to any patient and say nothing about yours.
Medical confidentiality and personal data: what must not be uploaded to a neural network
Under Article 13 of Federal Law No. 323-FZ, information about the fact of seeking medical care, health status, diagnosis, and other information obtained during examination and treatment constitute medical confidentiality. They must not be disclosed, including to those who learned them during training. A medical student during case curation is precisely such a person. Full name, date of birth, address, and policy number are classified as personal data under Law No. 152-FZ, and health information therein is classified as a special category.
The rule is simple: anonymize everything before uploading it to any neural network, including ours.
- Replace the full name with "Patient M." or simply "patient".
- Leave the age instead of the date of birth.
- Remove the address, phone number, policy number, SNILS, passport data, and medical record number.
- Do not specify recognizable workplaces and rare details that could be used to identify the patient.
- Do not take photos of documents and department discharge summaries. Cover anything that could identify a person on photos of your handwritten notes.
How not to get caught with someone else's work
Instructors usually check not with an "AI detector", but with questions at the defense: what did you hear during auscultation, why did you choose this drug, how does your diagnosis differ from the adjacent one in the differential series. A case history compiled from your curation notes is easy to defend. Nobody can defend a case history generated "from scratch" and unread.
- Reread each section and remove anything that the patient did not have.
- For each syndrome, name the data on which it is based.
- Find in the clinical guidelines where each line of treatment comes from.
- Check that complaints, examination, diagnosis, and follow-up notes do not contradict each other.
- Clarify the department's rules regarding AI usage with your instructor.
- Say the diagnosis justification out loud. On the platform, you can discuss the topic out loud with the AI examiner: it asks questions and, after the conversation, shows what you mentioned and what you missed.
If you need a report based on the case history, it is convenient to assemble slides from it: we explained how to do this in the article medical presentation via neural network. An overview of all educational tasks that AI handles is available in the material neural network for medical students.
Frequently asked questions
Can you write a medical case history using a neural network?
You can, if you managed the patient yourself and the neural network helps format the notes according to the required structure and cross-checks them with clinical guidelines. Before submission, the text must be reviewed, and you should be ready to defend every section.
Which neural network writes medical case histories in Russian?
In WhiteCoat Pro, a case history is compiled according to the department's protocol, checked against Russian Ministry of Health clinical guidelines and ICD-10 codes, and works in Russia without a VPN. Universal chatbots can generate text, but they do not know Russian clinical guidelines or your department's specific format.
Where can I find the medical case history template?
The official template is in the department's methodological guide. On the platform, you can choose one of the departmental templates added by participants or use the classic format.
How much does it cost to write a case history with AI?
Every account includes 2 case histories per month for free, and the «Maximum» plan includes 5 more. Plan details are available on the subscription page.
Will the instructor know the case history was written by AI?
This usually comes to light during the defense when questioned about the patient. If the case history is built on your own data and you have thoroughly worked through it, answering will be easy.
Can I upload a photo of a ward case history into the neural network?
No. This is a medical document containing personal data and information constituting medical confidentiality. Upload only your own anonymized notes.