What the person reports, scored the way it was validated.
PHQ-9, GAD-7, PCL-5, MDQ and the C-SSRS screener scored with their published rules, language markers from what the person says, and ICD-10-CM codes checked and grouped.
For research and clinician review. Not a diagnosis.
Clinical context, step by step.
0:57 with on-screen captions. Every number in it is the engine's output for the published sample.
Read the transcript
01 · What the person reports
Validated questionnaires, the person's own words, and the record: scored as published.
02 · Instruments
PHQ-9, GAD-7, PCL-5, MDQ and the C-SSRS screener are scored with their published rules.
Nine answers total 13: the moderate band, exactly as published.
03 · Safety first
Item 9 and any C-SSRS risk route to a clinician, whatever else the session shows.
Here item 9 is zero, so the priority stays routine.
04 · Language
From the transcript, words are counted per 100: first person, absolutist, negative affect, fillers.
English and Arabic, with the lexicons from Medera BH-ASR.
05 · The record
ICD-10-CM codes are checked and placed in their Chapter 5 block.
06 · Validation
An answer out of range is rejected with its exact path. Nothing is guessed.
Language markers need 20 words; C-SSRS follow-ups are required when question 2 is yes.
07 · One request
One request: answers, a transcript and codes in; scores, bands, markers and safety out.
Scores describe the questionnaire. The inference engine turns them into probabilities.
08 · Recap
Scored as validated, safety first, and nothing guessed.
Inputs, and only those.
Responses and transcripts are processed in memory and discarded when the response is sent.
- 01Item responses for PHQ-9, GAD-7, PCL-5, MDQ and the C-SSRS screener
- 02A transcript in English or Arabic
- 03ICD-10-CM diagnosis codes from the record
Published methods, in order.
Each step names the method behind it. The catalog lists the references for every measure.
- 01Validate every item
Each instrument must arrive complete and in range. An invalid answer is rejected with its exact path, never guessed.
- 02Score and band
Totals and severity bands exactly as published: Kroenke 2001 for the PHQ-9, Spitzer 2006 for the GAD-7, the National Center for PTSD rules for the PCL-5, Hirschfeld 2000 for the MDQ.
- 03Route safety first
Any positive PHQ-9 item 9 and any C-SSRS risk level go straight to a clinician, whatever else the session shows.
- 04Read the language
First-person singular, absolutist and negative-affect words, fillers and repetitions per 100 words, with BH-ASR's English and Arabic lexicons.
- 05Place the record
ICD-10-CM codes are checked for shape and grouped into their Chapter 5 blocks.
Scores, markers and safety.
Every measure links to its catalog entry: definition, method, quality gate, published associations and references.
| Instrument | Answers | Scoring | Source |
|---|---|---|---|
| PHQ-9 | 9 items, 0 to 3, plus difficulty | Total 0 to 27 in the published bands: minimal 0 to 4, mild 5 to 9, moderate 10 to 14, moderately severe 15 to 19, severe 20 to 27. Item 9 above 0 requires clinician review. | Kroenke 2001 |
| GAD-7 | 7 items, 0 to 3 | Total 0 to 21: minimal 0 to 4, mild 5 to 9, moderate 10 to 14, severe 15 to 21. | Spitzer 2006 |
| PCL-5 | 20 items, 0 to 4 | Total 0 to 80, positive at 33 or more, with the DSM-5 symptom-cluster rule reported alongside. | National Center for PTSD |
| MDQ | 13 yes or no, timing, impairment | Positive with 7 or more yes answers in the same period and moderate or serious problems. | Hirschfeld 2000 |
| C-SSRS screener | Questions 1 to 6 | Columbia SAFE-T risk level from the answers: low raises the priority to elevated, high to urgent. Risk and protective factors need clinician input and are never computed. | Posner 2011 |
| Biomarker | Unit | What it measures | Published associations |
|---|---|---|---|
| Language | |||
| Wordswords | count | Words spoken. | Anxiety↓SSchizophrenia↓MA |
| First-person singularfirstPersonSingularPer100 | /100 words | I, me, my, mine, myself (and Arabic equivalents) per 100 words.RDoC: Social processes | Depression↑MA |
| Absolutist wordsabsolutistPer100 | /100 words | Absolutist words (always, never, completely, nothing) per 100 words.RDoC: Negative valence systems | Depression↑SAnxiety↑SSuicide risk↑S |
| Negative-affect wordsnegativeAffectPer100 | /100 words | Negative-emotion words from a curated lexicon per 100 words.RDoC: Negative valence systems | Depression↑SAnxiety↑S |
| FillersfillerPer100 | /100 words | Fillers (um, uh) per 100 words; needs a verbatim transcript that keeps them. | Descriptive |
| RepetitionsrepetitionPer100 | /100 words | Immediate word repetitions per 100 words. | Descriptive |
Associations are group-level findings from the literature (↑ higher, ↓ lower; MA meta-analysis, R replicated, S single study), never a diagnosis. Compare a person with themselves over time.
Null, never a guess.
When the data cannot support a measure, the value is null and the response says why in quality.flags.
- Language markers need at least 20 words.
- C-SSRS questions 3 to 5 are required when question 2 is yes.
- Risk and protective factors in SAFE-T need clinician input and are never computed.
One request. Try it now.
Each example downloads the published sample and sends it with the public sandbox key. The sandbox takes synthetic data only.
# The published sample request (synthetic, no real person's data)
curl -s https://www.medera.info/api/v1/samples/clinical-context -o clinical-context.json
curl -s https://www.medera.info/api/v1/clinical-context/analyze \
-H "Authorization: Bearer mk_sandbox_medera_public" \
-H "Content-Type: application/json" \
--data @clinical-context.jsonSends the synthetic sample to the live sandbox with the public key and shows the full response. Nothing personal is sent and nothing is stored.
{
"object": "clinical_context.analysis",
"api_version": "v1",
"mode": "sandbox",
"engine": {
"name": "medera-inference",
"version": "1.0.0",
"vocal": "BH-ASR voice layer 1.2.0 (TypeScript port, number-for-number with the Python reference)"
},
"intended_use": "Measured biomarkers and within-person change for research and clinician review. Not a diagnosis and not a medical device.",
"instruments": {
"phq9": {
"instrument": "phq9",
"total": 13,
"band": "moderate",
"band_range": "10–14",
"item9": 0,
"item9_positive": false,
"difficulty": 2
},
"gad7": {
"instrument": "gad7",
"total": 9,
"band": "mild",
"band_range": "5–9"
},
"pcl5": null,
"mdq": null,
"cssrs": null
},
"language": {
"words": 57,
"firstPersonSingularPer100": 15.789,
"absolutistPer100": 3.509,
"negativeAffectPer100": 3.509,
"fillerPer100": 1.754,
"repetitionPer100": 0,
"counts": {
"first_person_singular": 9,
"absolutist": 2,
"negative_affect": 2,
"filler": 1,
"repetition": 0
},
"language": "en"
},
"biomarkers": {
"words": 57,
"firstPersonSingularPer100": 15.789,
"absolutistPer100": 3.509,
"negativeAffectPer100": 3.509,
"fillerPer100": 1.754,
"repetitionPer100": 0
},
"diagnoses": [
{
"code": "F32.1",
"valid_format": true,
"chapter5_block": "F30-F39",
"block_title": "Mood [affective] disorders"
}
],
"safety": {
"review_required": false,
"priority": "routine",
"reasons": []
}
}id and a created timestamp. Full referenceClinical context, answered.
Still wondering about something?
A real person will answer, usually the same day.
No. Scores and bands describe the questionnaire. The inference engine turns them into post-test probabilities with stated priors, and a clinician decides.
The response carries a safety block that requires clinician review, with a higher priority when the answer is more than half the days or nearly every day.
Better together, compared over time.
Send this modality with the others in one call to /v1/analyze, or pass the biomarkers to the inference engine.
Start measuring.
Sandbox today. Production under agreement.
Or make a first request with the sandbox key.