1
Which clinical use case?
- Select one high-value German workflow
- Define intended user, output and clinical claim
- Confirm the problem with target KOLs
VALIRAX
01 / 18
VALIRAX
BIOVENTURES
A proposed Clinical Autism AI platform for Germany
Approve €100k for Phase 1A validation
Board decisionPrepared for Serefin Health • September 2026
- PHASE 1AValidate & design
- PHASE 1BPilot
- PHASE 2Build & evidence
- PHASE 3Commercialize & scale
Why act
Germany cannot meet current autism diagnostic demand
Clinical AI could improve specialist productivity and patient throughput
1,300monthly requests
35-40appointments available
≈3%illustrative acceptance rate
2+yearsadult waiting time
AI can create specialist-equivalent capacity without proportional headcount growth.
Source: Bundesverband Autismus Deutschland, diagnostic access update, August 2026
Why it matters
Direct healthcare cost is measurable. The wider societal burden is substantially larger, but remains modeled.
Measured direct healthcare cost
€3,287
per person / year
≈ €274 / month
Largest reported components
19.8%Psychiatric inpatient
11.1%Pharmacotherapy
11.1%Occupational therapy
n=385 | ages 4-67
VISIBLE TIP|HEALTHCARE
BELOW THE WATERLINE|SOCIETAL COSTS
Disability & participationeducation & transitionproductivity & informal care
Modeled total
societal burden
Modeled, not official German autism-cost statistics
Investor takeaway: €274/month is the evidenced healthcare floor; the majority of the economic burden sits outside healthcare.
Source: Höfer et al. (2022), J Autism Dev Disord 52(2):540-552. Societal scenarios use German disability-spending anchors and explicit assumptions. Research cut-off: 13 Sep 2026.
Strategic thesis
Clinical value supports reimbursement; coordination and data strengthen adoption and defensibility
AUTISM INTELLIGENCE PLATFORM
The investable proposition is regulated clinical value, not a standalone navigation service.
Select a node to pause the cycle on it.
Proposed platform
One platform linking clinical decisions, care execution and outcomes.
Clinical AI creates medical value. Coordination converts recommendations into measurable follow-through.
LONGITUDINAL DATA LAYER • clinical, engagement and outcome dataLONGITUDINAL DATA LAYER • clinical, engagement and outcome data
Human oversight and clinical accountability remain integral to every regulated decision.
Board recommendation
Approve Phase 1A validation only.
It must resolve three questions before any pilot or build commitment.
2
Which product pathway?
- Separate non-clinical coordination from regulated functionality
- Map provider, payer and potential DiGA routes
- Translate claims into evidence requirements
3
Is a pilot executable?
- Identify target sites and workflow
- Define endpoints and data requirements
- Build a realistic next-phase budget
Outcome: a fact-based GO, PIVOT or STOP decision.
Scope Phase 1A
Define one investable German use case.
Phase 1A designs the pilot; Phase 1B would execute it.
Phase 1A objective
Define one priority clinical use case and design an investable German pilot.
01
Clinical · KOL
Validated target workflow, unmet need and user.
Confirm workflow, unmet need, adoption barriers and clinical success criteria
02
Product · Regulatory
Target product profile, claims and module boundaries.
Define intended purpose, users, outputs, module boundaries and likely pathway
03
Pilot design · Evidence plan
Pilot design, endpoints, sites and data needs.
Design sites, workflow, endpoints, data requirements and governance.
04
Commercial · Capital
Commercial route, next-phase budget and gates.
Test buyer, reimbursement logic, pricing hypothesis, and build/pilot budget
One target product profile aligns clinical value, regulation, evidence and economics
No pilot is executed in Phase 1A
Phase 1B objective
Execute German Clinical Use Case Pilot
Roles & decision rights
Serefin decides; Valirax validates and designs execution.
Responsibilities are explicit and the final investment decision remains with Serefin.
SEREFIN HEALTH
Decides- Provide platform, existing assets and product constraints
- Nominate decision owner and clinical/product contributors
- Decide intended product direction and acceptable risk
- Approve the GO / PIVOT / STOP outcome
VALIRAX BIOVENTURES
Validates & designs- Mobilize German KOL and pilot network
- Lead market-access, pathway and pilot design
- Integrate clinical, regulatory and commercial findings
- Deliver the investment recommendation and next-phase plan
Governance: agreed workplan, assumption log, decision register, and final board review
Investment & outputs
€100k buys a defined decision package
The proposed collaboration is limited to advisory and validation work.
Proposed investment
€100k
Phase 1A
Advisory + validation package
Follow-on capital requires a separate GO decision.
Core decision documents
- Clinical use-case recommendation
- Target product profile
- Regulatory / reimbursement route memo
- Commercial hypothesis
Next investment recommendation
- Integrated GO / PIVOT / STOP recommendation
- Scope, budget and milestones for the pilot/build phase
- Explicit assumptions requiring external specialist confirmation
WHY VALIRAX?
Valirax accelerates German market entry through:
- KOL access
- Clinical partnerships
- Digital health expertise
- Pilot execution capabilities
Valirax reduces execution risk and accelerates commercialization
Board resolution
Approve Phase 1A. Validate before committing.
Investment
Up to €100k
Consulting agreement
Phase 1A output
GOPIVOTSTOP
One use case and Phase 1B investment case
Not authorized
Pilot Build Partnership Exclusivity
Any such commitment requires a separate approval after Phase 1A.
Proposed
resolution
resolution
Approve expenditure of up to €100k for Phase 1A under a consulting agreement with Valirax Bioventures, and appoint joint decision owners. Phase 1A will return a GO, PIVOT or STOP recommendation. No later-phase commitment is authorized.
Collaboration roadmap
Phase 1A validates and designs; later phases execute
Every transition requires a separate scope, agreement and investment approval
VALIDATE & DESIGN
Consulting agreement- Validate clinical need
- Select use case and define product
- Design pilot and Phase 1B case
PILOT
Partnership agreement- Configure initial solution
- Execute selected-site pilot
- Test workflow and adoption
BUILD & EVIDENCE
Partnership agreement- Develop regulated modules
- Establish quality and governance
- Generate claim-aligned evidence
COMMERCIALIZE & SCALE
Partnership agreement- Launch commercially
- Expand reimbursement
- Replicate by market/channel
Approval today covers Phase 1A only. Pilot execution, product development and commercialization remain separate decisions.
Future partnership
A separate partnership agreement would govern Phase 1B and beyond
If Phase 1A results in GO, the parties would negotiate partnership terms before any pilot begins.
SCOPE & DECISIONS
- Joint roadmap and target markets
- Work packages, milestones and approval rights
- Budgets, change control and performance obligations
NEGOTIATE BEFORE PHASE 1B
IP, DATA & LICENSING
- Background IP remains with each party
- Phase 1A deliverables covered by consulting terms
- Future IP, data use, licensing and termination rights agreed separately
NEGOTIATE BEFORE PHASE 1B
COMMERCIAL MODEL
- Development funding and cost allocation
- Potential fees, licensing, royalties or revenue share
- Market, channel and commercialization responsibilities
NEGOTIATE BEFORE PHASE 1B
GOVERNANCE & RISK
- Steering committee and escalation
- Regulatory, quality, privacy and liability allocation
- Termination, transition and dispute provisions
NEGOTIATE BEFORE PHASE 1B
Phase 1A approval does NOT approve a partnership, exclusivity, future IP allocation, pilot execution or follow-on funding
Appendix
Evidence & regulatory guardrails
What is verified, what is modeled, and what requires specialist validation.
Research cut-off: 13 September 2026
Evidence appendix
Evidence strength varies materially by claim
Investor diligence should separate primary evidence from modeled assumptions and strategic hypotheses.
| Claim | Current support | Confidence | Required next step |
|---|---|---|---|
| Adult diagnostic bottleneck | National association update: 2+ years; lists often closed/restricted | HIGH | Validate at 5-10 target centres |
| Direct healthcare cost | Peer-reviewed German outpatient sample: €3,287/year | MEDIUM | Inflate to current year; test claims data |
| Autism population | Prevalence-based model; no single official national count | MEDIUM | Triangulate claims, registry and survey data |
| €10.9B societal burden | German anchors plus support/productivity assumptions | LOW-MED | Obtain autism-specific SGB IX and productivity data |
| DiGA reimbursement | Established pathway, but scope-dependent eligibility | MEDIUM | BfArM advice before freeze of intended purpose |
Evidence hierarchy:German authority / peer-reviewed German study>professional association>modeled proxy>strategic hypothesis
Regulatory appendix
Product claims determine the regulatory pathway
The same software architecture can face different obligations depending on intended purpose, outputs and users.
Examples
Scheduling, directory search, reminders, document exchange
Regulatory reading
Often outside MDR if no medical purpose or clinical decision influence. GDPR and security still apply.
Examples
Symptom management, patient-relevant process improvement, treatment support
Regulatory reading
May be MDSW. Potential DiGA fit only if Class I/IIa, primarily digital, patient-used and positive care effect is demonstrated.
Examples
Individual diagnosis, triage, treatment recommendation or material clinical decision support
Regulatory reading
Likely MDSW; Rule 11 often starts at Class IIa and can rise with potential harm. Notified-body and AI Act analysis may apply.
Guardrail: keep coordination outputs non-clinical unless the regulated module, quality system, evidence and human oversight are ready.
Sources: MDCG 2019-11 Rev.1 (June 2025); BfArM DiGA Guide v3.6 (10 Dec 2025); EU AI Act Article 6 and Commission draft high-risk guidance (19 May 2026). Not legal or regulatory advice.
Reimbursement appendix
DiGA is a design constraint, not a label applied at launch
Treat eligibility as an early product architecture decision.
12345
MEDICAL DEVICE?
Clear medical intended purpose and CE marking under MDR.
ELIGIBLE CLASS?
DiGA entry is limited to eligible Class I or IIa devices.
PATIENT-USED?
Main digital function must be used by or jointly with the insured person.
POSITIVE CARE EFFECT?
Comparative evidence of medical benefit or patient-relevant structural/procedural improvement.
SECURE & INTEROPERABLE?
BfArM data protection, information security, usability and interoperability requirements.
Key caveat
for Serefin
for Serefin
A clinician-facing diagnostic engine may be commercially attractive through provider procurement but may not itself satisfy DiGA’s patient-use logic. A modular product strategy may be required.
Primary source: BfArM DiGA Guide v3.6, 10 Dec 2025. Eligibility and evidence should be confirmed through formal BfArM advice and regulatory counsel.
Risk appendix
Clinical AI requires a safety and evidence system, not only a model
Human oversight is necessary but does not replace validation, governance or accountability.
CLINICAL SAFETY
Define intended use, failure modes, escalation, contraindications and clinician accountability.
BIAS & GENERALISABILITY
Validate by age, support needs, language, sex, comorbidity and care setting; monitor subgroup performance.
DATA PROTECTION
Health data are special-category data under GDPR Article 9; establish Article 6 basis plus Article 9 condition, minimisation and governance.
AI GOVERNANCE
Classify under the AI Act; document data governance, transparency, human oversight, performance and post-market monitoring as applicable.
CYBERSECURITY
Secure development lifecycle, access controls, logging, incident response, vulnerability management and supplier assurance.
EVIDENCE
Prospective comparative evidence aligned to the exact claim and reimbursement endpoint; avoid validating a different product version.
Sources: GDPR Regulation (EU) 2016/679, Article 9; EU AI Act Article 6; MDCG 2019-11 Rev.1. Specialist legal, regulatory, clinical and cybersecurity review required.
Sources
Core evidence register
Primary and near-primary sources used to qualify investor claims. Research cut-off: 13 September 2026.
| Area | Source | Supports | Link |
|---|---|---|---|
| AUTISM ACCESS | autismus Deutschland, “Diagnostik,” status 18 Aug 2026 | 2+ year adult waits; widespread closed/restricted waiting lists | SOURCE |
| HEALTHCARE COST | Höfer et al., J Autism Dev Disord 52, 540-552 (2022) | €3,287 average annual service cost in n=385 outpatient sample | SOURCE |
| DISABILITY SUPPORT | Destatis, Eingliederungshilfe 2024 | €28.7B net spending; 1.029M recipients across all disabilities | SOURCE |
| DIGA | BfArM DiGA Guide v3.6, 10 Dec 2025 | Eligibility, application and evidence framework | SOURCEBfArM overview |
| MDR SOFTWARE | MDCG 2019-11 Rev.1, June 2025 | Qualification and Rule 11 classification of medical device software | SOURCEUpdate note |
| AI ACT | European Commission Article 6 / draft guidance, 19 May 2026 | High-risk classification logic; draft guidance is not final | SOURCE |
| DATA PROTECTION | Regulation (EU) 2016/679, Article 9 | Health data are special-category personal data | SOURCESummary |
All market-size and societal-cost totals remain estimates unless explicitly labeled as official statistics. URLs embedded in SOURCE buttons.