CareEZ —
Compared to alternatives
An honest, evidence-based comparison. Decision-makers ask how CareEZ differs from existing dysphagia tools. This page gives a direct answer — including where we fall short today.
Intended for: RCHE administrators, HKEX panel reviewers, NGO directors, SLP researchers, and procurement teams.
What each tool actually does
We have tried to be accurate and fair to each category. Sources are cited where available. If a competitor's team believes a cell is wrong, please email [email protected] — we will update promptly.
| Feature | CareEZ | SimplyThick / Nestlé ThickenUp | Free HKCSS chatbots | Smartphone IDDSI apps |
|---|---|---|---|---|
| AI-powered IDDSI classification | ✅ YesVision + text + voice multimodal | ❌ NoProduct label guidance only | ❌ NoBasic text Q&A | ⚠️ BasicSingle-modal text/image only |
| Multilingual support | ✅ 8 languagesEN / zh-HK / tl / id / vi / th / es / zh-CN | ⚠️ EN + 中simplymedical.com; Resource Hub | ⚠️ EN + 中hkcss.org.hk resources | ⚠️ VariesTypically EN only |
| FDH-language support (tl / id) | ✅ YesTagalog + Bahasa Indonesia natively | ❌ | ❌ | ❌ |
| Live open API | ✅ Apache-2.0api.seniordeli.com — public, documented | ❌ | ❌ | ❌ |
| Clinical-safety bias (over-refer on ambiguity) | ✅ ExplicitHard-coded in model prompt; documented | N/AProduct, not classifier | ⚠️ Unclear | ⚠️ Varies |
| Co-governed public asset | ✅ YesHKU co-development + s.88 charity trust | ❌ ProprietaryNestlé / SimplyThick Inc. | ⚠️ HKCSS-only | ❌ Proprietary |
| Free at point of use for carers | ✅ Always free | ❌ Product costThickener purchase required | ✅ Free | ⚠️ VariesSome paid apps |
| HKCSS 護食標準 alignment | ✅ ExplicitMapped to 4-level 照護食 scale | ❌IDDSI-only, no 護食標準 mapping | ✅ Direct sourceHKCSS publishes the standard | ⚠️ Varies |
| Real food image classification | ✅ Claude VisionPhoto → IDDSI level in <3 sec | ❌ | ❌ | ❌ |
| Voice biomarker screening | ⚠️ Research-stagedY1–Y2 roadmap; not yet diagnostic | ❌ | ❌ | ❌ |
| WhatsApp / WeChat integration | ⚠️ SkeletonMVP live; expanding Q3 2026 | ❌ | ❌ | ❌ |
| Thickener dosage calculator | ✅ YesProduct-agnostic; brand-neutral | ✅ YesFor own product line only | ❌ | ⚠️ Some apps |
✅ = available and verified | ❌ = not available | ⚠️ = partial or uncertain | N/A = not applicable to product category.
Last reviewed: May 2026. Corrections welcome at [email protected].
How each category is different — and complementary
These are not competitor takedowns. Each tool below solves a real problem. The distinctions matter for procurement decisions.
vs Texture modifier products SimplyThick & Nestlé ThickenUp
These are excellent, clinically validated products — we recommend them to carers and facilities. They solve the "what thickener to use and how much" problem for safe liquid consumption.
CareEZ solves a different problem: "is what I am about to serve at the right consistency for this resident?" — before a meal is plated, using a photo of real food.
The two are complementary. A carer can use ThickenUp to prepare a drink, then photograph it with CareEZ to confirm the texture level before serving.
SimplyThick: simplythick.com ↗ | Nestlé ThickenUp: nestlehealthscience.com ↗
vs Standard-setting bodies HKCSS chatbots & awareness sites
HKCSS publishes and governs the 護食標準 (照護食). They are the authoritative source — we build on their standard, not against it.
HKCSS's digital tools are designed for Chinese-speaking clinicians and families. CareEZ extends reach to FDHs (Tagalog, Bahasa Indonesia), Vietnamese, Thai, and Spanish-speaking community carers who are not served by existing resources.
We do not replicate HKCSS's work. We extend its adoption to frontline non-clinician staff who have never been served by a dysphagia tool in their own language.
HKCSS 護食標準: hkcss.org.hk ↗
vs App-based tools Smartphone IDDSI apps
Several apps (primarily in Australia and the UK) let clinicians look up IDDSI levels via text search or reference guides. They are useful reference tools for trained SLPs.
CareEZ differs in three ways: multimodal (photograph real food instead of text search), multilingual (8 languages, not just English), and open API (any EHR or facility system can query it).
Most existing apps are designed for SLPs, not untrained family carers or FDHs. CareEZ targets the last-mile untrained carer — the person actually serving the meal.
vs General AI assistants ChatGPT / general LLMs
General AI assistants have been used ad-hoc for IDDSI questions. They can answer text questions but have no clinical-safety bias, no HKCSS 護食標準 mapping, no image classification for food, and no dosage calculator.
CareEZ is purpose-built: the model is constrained to over-refer on ambiguity (never under-classify), the output maps to Hong Kong clinical standards, and the system is co-governed by a university clinical team, not a commercial AI vendor.
Honest limitations — today
We believe transparent disclosure of limitations is a clinical obligation, not just a marketing posture. These are our current known gaps.
- Vision accuracy: 4 / 8 IDDSI levels under baseline test (50%) — classification of pureed and minced textures remains unreliable at current model scale. We are addressing this via a fine-tuning run in Y1 Q3 2026 using a proprietary annotated image dataset.
- Voice biomarker is research-staged, not yet diagnostic — wet-voice detection has shown promise in pilot recordings but has not been validated in a clinical trial. It will not be surfaced in production until peer-reviewed validation is complete (target: Y2 Q1).
- No regulatory clearance (FDA / NMPA / MDCO) — CareEZ is positioned as a triage and decision-support tool, not a medical device. It does not make diagnoses. Clinical decisions remain with qualified SLPs and physicians. We do not plan to seek device clearance in Y1.
- WhatsApp / WeChat integration is skeleton-only — the API backbone is live; the consumer-facing chatbot channels are in MVP, not production. Rollout targeted Q3 2026.
- No published peer-reviewed outcome data yet — we have pilot feedback from partner RCHEs but have not yet published clinical outcome studies. A partnership with HKU is structured to produce this data in Y1–Y2.
Free for all carers. Always.
CareEZ is free at point of use for family carers, FDHs, and frontline care staff. Institutional licensing for RCHEs and NGOs covers operations and enables priority support, API quotas, and custom integrations. See full pricing at careez.org/pricing.