CSE for Health
CSE for Health is a cloud natural language processing system that codes hospital, outpatient and emergency reports into ICD-10 for hospitals and research centers. Coding is supervised: staff review every code falling below the confidence threshold.
What is CSE for Health?
CSE for Health is the commercial name of codingSuite Enterprise, a natural language processing system for the automatic coding of clinical reports and documents. It is published by Words for Knowledge Information Technology (W4Kit), a Barcelona company that describes a team of software engineers, linguists and NLP specialists and claims more than ten years of activity in AI-driven clinical coding.
By default the system codes into ICD-10, the World Health Organization catalog of roughly 140,000 entries. It can also be trained on ICD-11, on SNOMED CT, and on subsets or specialty extensions, and several catalogs can be used inside a single document on terms agreed with each client. Three document perimeters are covered — inpatient stays, the specialized outpatient area covering external consultations and day hospital, and emergency care — and the vendor states that this perimeter can be extended without limit.
The coding is supervised, not fully automatic. The software assigns a confidence score to each code; codes above the threshold set by the hospital return straight to the hospital information system, while everything below it goes to the clinical documentation team for manual review. The NLP layer is built to handle synonyms, acronyms, spelling and typographic errors, compound words, regional variants and language switches inside the same text. The stated quality objective is automation and accuracy above 90%, and the home page cites an effectiveness rate of more than 90%.
Delivery is SaaS: a cloud system reached over HTTPS from a browser, with no installation and no infrastructure cost for the institution. Integration with the HIS runs through web services, with documents exchanged in batches or online depending on the center's preference. Quality, efficiency and effectiveness are exposed through control dashboards the vendor presents explicitly as an anti-black-box measure.
The rationale behind the product is that 75% of health data is unstructured. It has been deployed at Hospital Clínic de Barcelona since 2013 and at Fundació Puigvert since 2010, where it started under the name CodingSuite; the site footer also lists Universitat de Barcelona among its collaborations.
What it does
- Automatically code inpatient discharge summaries, emergency reports and outpatient consultation notes into ICD-10
- Attach a confidence score to every code and route only those below the institution's chosen threshold to manual review
- Learn from the documentation team's corrections and improve coding continuously
- Turn unstructured clinical text into structured, queryable data
- Feed the CMBD return to the health administration within the regulatory deadline
- Exchange documents and results with the hospital information system through web services, in batches or online
- Track quality, efficiency and effectiveness through anti-black-box control dashboards
When to use CSE for Health / When not to
A quick filter to help you decide if CSE for Health is the right fit.
When to use CSE for Health
- Hospitals with an in-house clinical documentation team that spends most of its time on repetitive, high-volume coding and wants to refocus that effort on complex cases.
- Hospitals that currently outsource their coding and want to bring the whole volume back in-house.
- Health information management departments under pressure to file a complete, on-time CMBD return to the health administration.
- Research centers that need clinical documents turned into structured, queryable data for case-mix analysis and studies.
- Coding departments running a catalog migration from ICD-9 to ICD-10, as at Hospital Clínic in 2017 and Fundació Puigvert in 2019, or absorbing a rise in volume without adding headcount.
When not to use CSE for Health
- Individual clinicians or freelance coders: the service is deployed at institution level and integrated with the hospital information system, not sold to a single user.
- Buyers who want to evaluate a tool self-service: there is no published price, no online sign-up and no trial, and every entry point runs through a sales conversation.
- Teams looking for a mobile app, a desktop client or a browser extension: CSE for Health exists only as a web service consumed by the hospital's own systems.
- Anyone who needs coding outside healthcare, or who expects the software to write the clinical report: it works on clinical catalogs only, and it codes documents that have already been produced rather than drafting them.
- Organizations that need English-language documentation or references beyond Catalonia: the site and its materials exist only in Spanish and Catalan, and the published references cover two Barcelona institutions.
How to use CSE for Health
A typical end-to-end flow, from setup to results.
- Request a demonstration through the contact page: there is no online sign-up and no self-service entry point.
- Agree the document templates with the vendor before go-live, section by section and for each document type the center uses.
- Choose the starting point: train on a corpus of already-coded documents, or start from zero using the fallback coder built on Hospital Clínic's experience.
- Set the confidence threshold above which codes are returned automatically and below which they go to human review.
- Connect CSE to the hospital information system through web services, and decide whether documents travel in batches or online, and at what frequency.
- Have medical staff write reports as usual in the hospital's existing system (SAP, HCIS, SAVAC or equivalent).
- Let the documents synchronize with CSE: inpatient, emergency and outpatient records.
- CSE processes each report, codes it against the specified catalog and returns to the HIS every code whose confidence exceeds the threshold.
- The documentation team reviews and corrects the codes that fell below the threshold, from a browser, with nothing to install.
- The system learns from those corrections and improves subsequent coding; periodic retraining and follow-up meetings keep the loop running.
Pros & Cons
Pros
- Long, verifiable hospital references: Fundació Puigvert since 2010, Hospital Clínic de Barcelona since 2013, and 100% of the Clínic's reports covered since 2019
- Published accuracy figures rather than vague claims: 94% at the Clínic in 2016 and close to 98% at Puigvert in 2013, against a current stated objective above 90%
- The confidence threshold is set by the institution, so each center arbitrates quality against review workload itself
- A learning loop fed by the center's own documentation staff, backed by a multidisciplinary team of engineers, linguists and NLP specialists and periodic follow-up meetings
- Anti-black-box dashboards: non-opacity is stated as an explicit commitment rather than left implicit
- No installation and no infrastructure cost for the institution, on an annual contract with no minimum-term clause
- Privacy posture stated up front: the templates sent for coding carry no patient personal data, and client data is committed never to be used commercially or for other clients' benefit
Cons
- No public pricing of any kind, not even an order of magnitude, and no free plan or trial is mentioned: the only route in is a demonstration request handled by sales
- No API documentation is published, even though integration is explicitly done through web services
- No data processing agreement and no subprocessor list for the clinical processing chain, and no security certification such as ISO 27001, ENS or SOC 2 is claimed
- No hosting country or region is stated for client data, and no retention period is published in figures: the policy points to criteria rather than to a deadline
- Site and materials exist only in Spanish and Catalan, and the multilingual coding claim names no language at all
- Public references are limited to two institutions, both in Barcelona; there is no blog, no social media presence and no public product documentation
- Maintenance signals are weak: the footer copyright still reads 2024, and the demonstration request page listed in the sitemap returns a 404
Pricing & Plans
No price is published. The complete sitemap contains no pricing page, and no rate, tier or order of magnitude appears anywhere on the site; no free plan and no free trial are mentioned either. Commercial terms are set through a sales conversation, the only entry point offered being the demonstration request. The one commercial condition stated publicly is the engagement model: annual contracting, with no minimum-term clause. Infrastructure is described as carrying no cost for the institution, since the service runs in the cloud and an internet connection is sufficient. The use of several coding catalogs within a single document is left to be agreed with each client.
Data, GDPR & hosting
A consolidated view of how CSE for Health handles your data.
GDPR overview
GDPR implementation is explicit rather than implied. The privacy policy is built on Regulation (EU) 2016/679, Spain's LOPDGDD 3/2018 and the LSSICE 34/2002, and states that the information is provided under Art. 13 GDPR and Art. 11 LOPDGDD. Data subject rights are enumerated: access, rectification, erasure, objection, restriction, portability and withdrawal of consent, exercised by post or email to the controller with a copy of a national ID or equivalent document. The supervisory authority is named as the Agencia Española de Protección de Datos. Three things are missing: no data protection officer is named, no Art. 28 data processing agreement is published or even mentioned, and the policy carries no effective date, no revision date, no reference to international transfers and no standard contractual clauses. No Art. 27 EU representative is designated, which is consistent with a publisher established in Spain.
Who owns the data?
The privacy policy names WORDS FOR KNOWLEDGE INFORMATION TECHNOLOGY, S.L. (CIF B65268765) as the sole data controller: only the controller receives and processes the personal data, with no transfer to third parties except where a law compels disclosure. The About page adds a commitment that client data will never be used for commercial purposes or for the benefit of other clients, and the product FAQ states that the templates sent for coding carry only the fields of the document needed to code, with no personal information about the patient. One scope limit matters: the published policy addresses website visitors and does not separately set out ownership of the clinical documents the product processes.
Reuse rights
The privacy policy lists its purposes as performing the contracted services, managing email exchanges and, with consent only, commercial outreach and newsletters, on the legal bases of GDPR Art. 6.1.b (contract), 6.1.a (consent) and 6.1.f (legitimate interest), plus Art. 20 of the Spanish LSSICE. Nothing in the published texts restricts what a hospital may do with the codes returned to its own systems, but no reuse or licensing clause is stated either way, so the point is unaddressed rather than settled. In the other direction, the vendor does reuse client input: CSE learns from the corrections the center's documentation staff make during review, and periodic retraining is presented as part of the service. No opt-out from that learning is documented anywhere on the site. Google Drive and Google Analytics, both provided by Google, Inc., are named as third-party tools used to manage user data, and complaints may be filed with the Spanish data protection authority, the AEPD.
Data retention & training
Hosting summary
Very little is published. The site describes a cloud system reached over the internet in HTTPS, with no infrastructure cost for the center, and that is the extent of it. No hosting country, no region, no cloud provider and no data center is named for client data. There is no mention of international transfers, no standard contractual clauses, and no processing agreement setting out where clinical documents are handled. One factual data point exists and must not be over-read: the marketing website itself resolves to 213.32.104.224, an OVH SAS address in Gravelines, France. That describes where the brochure site is served. It says nothing about where the coding platform runs or where client documents are stored, and the two should not be conflated. The only scoping element the vendor offers is functional rather than geographic: the FAQ states that the templates sent for coding carry only the fields of the document needed to code, and no personal information about the patient. For a hospital assessing this product, the hosting jurisdiction is a question to put to the vendor directly.
Things to keep in mind
Risks and trade-offs to weigh before adopting CSE for Health.
- The site contradicts itself on the publisher's identity: every footer credits W4Kit, S.L. while the legal notice and the privacy policy name WORDS FOR KNOWLEDGE INFORMATION TECHNOLOGY, S.L., the registered corporate name. Two addresses coexist as well, the registered office in Montgat and the contact address at Castillejos 265 in Barcelona, and the controller's contact email sits on a domain separate from the product (w4kit@w4kit.com).
- A second internal contradiction concerns tracking: the cookie policy states that only a few first-party cookies are used, while the privacy policy declares the use of Google Analytics.
- The system learns from the corrections your documentation staff make, and no opt-out from that learning is documented. The vendor commits not to use client data commercially or for the benefit of other clients, but that commitment lives on a web page rather than in a published contract.
- No hosting country is stated for the clinical processing chain, no Art. 28 data processing agreement is published, and no subprocessor list exists for clinical documents, which is notable when the publisher acts as a processor on a hospital's behalf.
- No retention period is published in figures: the policy points to criteria only, and carries neither an effective date nor a revision date.
- The accuracy rates quoted, 94% and close to 98%, date from 2016 and 2013 and concern two specific institutions. Reading them as a current guarantee for your own document mix would be a mistake; the commitment stated today is above 90%.
- The confidence threshold is a governance decision, not merely a setting: everything above it returns to the hospital information system without a human reading it. Set it low to save review time and coded data quality rests entirely on the model, while the anti-black-box dashboards are the only published means of noticing a drift, and no methodology behind them is documented publicly.
Setup & Integrations
Technical difficulty
Two very different levels. For the end user, effectively zero: no installation, a browser and the application's home page are enough. For the institution, this is an integration project: connecting CSE to the hospital information system through web services, defining document templates section by section for each report type, choosing the confidence threshold, and setting the exchange mode, batch or online, and its frequency. Initial training on already-coded documents is optional, since a fallback coder allows a start from zero. The vendor says it adapts to each center's infrastructure constraints and holds periodic follow-up meetings. Not a self-service rollout.
Deployment
Integrations
Behind CSE for Health
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What exactly is delivered?
How is it installed?
Which coding catalog does it use?
Which documents does it cover?
Do we need an existing training corpus to start?
How do documents move between the hospital and CSE?
Is patient personal data sent to CSE?
How accurate is it?
What does it cost?
Who publishes the tool?
Should you pick CSE for Health?
CSE for Health is a narrow, single-purpose product, and that is its strength. Automatic clinical coding is all it does, and it has been doing it in production for a long time: Fundació Puigvert since 2010 under the name CodingSuite, Hospital Clínic de Barcelona since 2013, and across the whole hospital since 2019.
The important nuance is that coding stays supervised. The confidence score and the threshold set by the institution make human review structural rather than transitional, which is reassuring clinically, but it also means the tool reduces coding effort rather than removing it. The accuracy figures the site publishes, 94% at the Clínic in 2016 and close to 98% at Puigvert in 2013, are milestones on project timelines for two specific centers, not a guaranteed current performance; the commitment displayed today is above 90%.
Against that, the commercial and technical opacity is complete. There is no published price, no trial, no public product documentation and no API reference, even though integration explicitly runs on web services. Nothing is said about where client data is hosted, no processing agreement is offered, and no security certification is claimed, which is unusual reticence for a vendor handling clinical documents on a hospital's behalf. The public footprint is narrow too: Catalonia, Spanish and Catalan only, no blog, no social media, a footer copyright still reading 2024.
The tool is a credible fit for hospitals that must produce a high-quality CMBD at volume and can run an integration project alongside their HIS team. Before signing, ask for the hosting jurisdiction, an Art. 28 processing agreement, a retention period in figures, and references outside Barcelona.
- Choosing a selection results in a full page refresh.
- Opens in a new window.