The plan: start with in-hospital insurance compliance, then the whole care chain
We plan to start where demand is strongest, risk lowest and validation fastest: in-hospital insurance compliance and record QC. From there we would extend to hospitals, pharma, insurers, pharmacies, health management and families. By design, each solution is a digital worker that completes a whole task, acts within its A0–A4 authority, attaches an evidence chain and leaves the final call to a person.
Product previewKunXi Medical LLM is at the design and concept stage
This section describes what we plan to build and how we intend to do it — not finished products or results. Product forms, model sizes, technical approaches, features, interfaces, metrics, scenarios and timelines are design concepts that have not been built or validated; they may be adjusted or changed as development, validation and regulatory requirements evolve. Nothing here is a commitment, an offer or a claim of clinical effectiveness.
All rights reserved. No reposting, adaptation or commercial use without written permission.Read the copyright notice
07Follow-up79–100Follow-up plan and alertsA2 scheduling
Step 63: compliance check before the settlement list is uploaded
62steps done
2re-plans after backtracking
1low-confidence hand-over
3A3 actions approved
Planned first scenario
Planned first, in-hospital insurance compliance and record QC: the highest score on six criteria
NoticeThis module is a design concept, not yet built, and may change.
Five candidate scenarios scored on demand, clinical risk, data availability, validation time, ability to pay and delivery difficulty (5 = best). The scores are KunXi’s judgement, not measured results, and will be updated with future pilot data.
← Swipe to see the full table →
Candidate
Demand
Clinical risk
Data access
Validation time
Ability to pay
Delivery
Total
In-hospital insurance compliance & record QC (planned first)
5
5
4
5
3
3
25
Pharma real-world evidence (RWE)
4
4
2
3
5
3
21
Insurance underwriting & claims
4
4
2
4
4
3
21
To C chronic-care companion
4
3
3
3
2
4
19
Imaging / diagnostic aid
4
1
3
1
3
3
15
Note: for clinical risk and delivery, lower risk and easier delivery score higher; data access scores full marks when no regional authorisation is needed; total = sum of the six (out of 30).
Why this first
CNY 34.2bninsurance funds recovered nationwide in 2025; compliance pressure reaches every record
95%+of short-stay cases are paid by diagnosis group; coding and grouping errors hit hospital income directly
15,500national insurance rule points can serve as automatic verifiers: checkable and auditable
Non-diagnosticdocumentation, QC and compliance prompts only; expected not to be regulated as a medical device (subject to classification; to be confirmed)
SourcesNational Healthcare Security Administration 2025 statistical bulletin (CNY 34.2bn recovered); Xinhua 2026-03-16 (95%+ of short-stay cases paid by disease group); NHSA (15,500 rule points, 2025); NMPA classification guidance for AI medical software (2021); scores are KunXi’s judgement
Planned roadmap
The plan: hospitals first, then pharma and insurers, imaging later
NoticeThis module is a design concept, not yet built, and may change.
Planned firstIn-hospital insurance compliance & record QC
Planned on-premises, data stays in the hospital, no regional data authorisation needed; documentation, QC and compliance prompts only, no diagnosis.
Second wavePharma RWE · insurance claims
Would start once data authorisation is in place; the chronic-care companion would be validated at small scale in parallel.
LaterImaging diagnostic aids
Only once the Class III device registration path is clear, with clinical trials as device rules require.
Solutions by customer
Six customer groups, each with a planned digital worker that does the whole job
NoticeThis module is a design concept, not yet built, and may change.
Each solution concept spells out the problem, the steps the digital worker would take, what it would deliver, who signs off, and which core technology would sit behind it.
Hospitals
Insurance compliance, record QC and the inpatient journey
NoticeThis module is a design concept, not yet built, and may change.
The problem
Clawbacks and denials are rising; manual self-checks cover little and miss issues
Doctors carry heavy documentation; coding errors change DRG / DIP groups
What the digital worker would do
Insurance Compliance Officer (planned first)Reads the record before discharge, checks 15,500 rules, verifies the DRG group~30 steps
Real-world evidence pack · SGLT2 inhibitors and kidney outcomes in CKD 3–4 (illustration)Concept illustration · not a real product screen · simulated data
Cohort build
Regional populationkept 100%
Type 2 diabeteskept ~9%
CKD stage 3–4kept ~18%
New-user designkept ~31%
After propensity matchingkept ~62%
Composite kidney endpoint · event-free survival
HR 0.72 (95% CI 0.61–0.85)Illustrative result · to be checked against the SAP
Treated
Comparator
x: months
Data lineage and QC
Event-level timestamps98.6%
Traceable to source text100%
Key-variable missingness2.3%
Code-mapping agreement99.1%
SAP pre-reviewDone
Deliverable: evidence pack = protocol + SAP + analysis dataset + QC report + lineage list
Traceableevent timestamps + source location
Reproduciblea third party re-runs it from the SAP
Reviewablemapped to CDE guidance
Insurers
Underwriting, claims and pricing people with conditions
NoticeThis module is a design concept, not yet built, and may change.
The problem
Manual claims are slow; checking records and pre-existing conditions takes time
Risk for people with conditions is hard to assess and price
What the digital worker would do
Underwriting & Claims AgentReads records and pre-existing conditions, checks terms and liability, recommends underwriting and claim decisions~20 steps
Counterfactual risk simulationCompares future outcome distributions under different interventions to stratify risk
Planned output
Claim recommendation + evidence chain
Governance and sign-off
A3: recommendations carry source evidence; a claims handler signs off
KunXi HealthConcept illustration · not a real product screen · simulated dataGood morning, Mrs Wang · hypertension / type 2 diabetesFamily member · health manager Ms Li
Blood pressure todayOn target
132/84mmHg
Systolic, last 7 days
Medication · 08:00
Amlodipine 5 mg · metformin 0.5 gTaken
AI report reading (multimodal)
Chest CT
Ground-glass nodule, right upper lobe
About 6 mm, low risk
Recheck in 12 months
A1 draftSent after a partner doctor reviews it
Evidence: imaging report · lung-nodule guideline
Ask: should I worry about this nodule?
HomeRecordsConsultMe
Note: KunXi Medical LLM is a product preview at the design and concept stage. Unless a public study is cited, the products, capabilities, thresholds, metrics and timelines in this section are KunXi design targets, concepts or estimates — not achieved or measured results — and may change; product screens are concept illustrations with simulated data. By design, the agents only make suggestions; clinical decisions are made by doctors.
Build it with us
Explore putting a medical LLM to work on real clinical tasks
KunXi Medical LLM is at the design and concept stage. Hospitals, pharma companies, insurers, pharmacies and health-management providers who are interested are welcome to talk to us and assess data, workflow and the validation path together.
