{"id":"58b9d642-25fb-4fbf-bcae-18bac198133f","arxiv_id":"2412.09223","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"CSSDH is an OWL ontology combining ISO ContSys and WHO social determinants of health to support SDH data capture in continuity of care.","lead":"The paper builds CSSDH, an OWL ontology that adds social determinants of health concepts to the ISO continuity-of-care model ContSys. It is a candidate information model for capturing patients' social context in electronic health records, though it has not yet been tested on real data.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Section 4 evaluation is internally inconsistent: CQ1 is mapped to the wrong properties and the CQ2 SPARQL query does not enforce the crowding condition, so the published evidence does not support the interoperability claim.","rationale":"I read the paper in good faith: the authors have built a nontrivial artifact (171 classes, 141 object properties, 210 data properties) and run HermiT, a DL query, and an OOPS! scan. Those actions support the modest claim that a taxonomy was constructed and is consistent as far as those checks reach. The reader's weakest assumption — that two author-written competency questions plus internal consistency checks are insufficient evidence for the interoperability claim — is correct. My stress test makes that concern more concrete: the two published competency-question evaluations are not merely insufficient; they are internally mismatched. CQ1 is explicitly assigned the wrong properties, and the CQ2 SPARQL query lacks the required filter for crowding. This means the paper's own evaluation does not establish the narrower claim that the model answers its two stated competency questions. I do not see this as grounds for rejecting the underlying proposal outright, because the issue is fixable: release the OWL artifact, correct the competency-question mappings, and run a clean query test against a minimal dataset. Therefore I keep the reader's CONDITIONAL verdict unchanged, but I would make the release of the artifact and a corrected query evaluation explicit conditions of acceptance. No ad hominem is intended; the critique is about the argument and evidence, not the authors.","tokens_in":4485,"tokens_out":5163,"duration_ms":53725,"concrete_test":"Request the CSSDH OWL file and run the following minimal test: create four patient instances — (1) lives in a low-income area only, (2) laid off only, (3) crowded home only, (4) laid off and crowded home. Encode each scenario using the CSSDH properties, then execute CQ1 and CQ2 as SPARQL queries. CQ1 should return instance 1 (and any other low-income-area patients); CQ2 should return only instance 4. If the ontology has no property or class for 'lives in a low-income area,' or if the CQ2 query returns instance 2, then Section 4's evaluation fails and the claimed interoperability support is not demonstrated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that CSSDH is a formal ontological model that both captures social determinants of health and supports data interoperability. To establish this, the paper must at least show that the implemented model answers its own competency questions. Section 3.2 lists CQ1 ('patients lives in a low-income area') and CQ2 ('laid off from their job and whose houses are also crowded'). Section 4 then states: 'To address CQ1, we associated the data properties LayOffFromJob and CrowdingAtHome with the class subjectOfCare' — these properties do not represent low-income area, so CQ1 is not actually addressed. The displayed SPARQL query is for CQ2 but uses OPTIONAL for CrowdingAtHome without any FILTER, meaning a laid-off patient with no crowding data would be returned, contradicting the conjunctive intent of CQ2. Additionally, no OWL file is provided, so the DL query in Figure 3 is not independently reproducible. Because the two published checks are the only evidence that the model 'is logically sound in retrieving information,' this internal mismatch undercuts the narrower query-answering claim, and with it the broader interoperability claim. The failure is in the evidence, not necessarily in the taxonomy: a corrected implementation and a released artifact could still resolve it, but as written the paper does not demonstrate the central assertion.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes CSSDH, an OWL ontology that combines ISO/DIS 13940:2024 ContSys with WHO Social Determinants of Health concepts, with the stated aim of capturing social determinants of health in continuity-of-care data and enabling interoperability across EHR systems. The methodology section describes reuse of SOHO, the Social Prescribing Ontology, ICD-11 chapter 24, and Gravity project value sets, reporting 171 classes, 141 object properties, and 210 data properties, of which 171 are SDH-related. Evaluation consists of a HermiT consistency check, one DL query shown in Figure 3, a SPARQL sketch for two competency questions (CQ1 and CQ2), and an oops! pitfall scan. The authors state that future work will test the model with NHS Scotland, HSE Ireland, MIMIC-III, and Recoletas Red Hospital datasets.","tokens_in":4715,"tokens_out":3946,"duration_ms":37726,"significance":"If the central claim were fully supported, the contribution would be a practically relevant integrated ontology: it explicitly reuses established standards and vocabularies (ContSys, SOHO, ICD-11, Gravity), states concrete competency questions, and reports machine-checked consistency via HermiT and an oops! scan. Those are genuine strengths. However, the published evidence currently establishes only that a model of this design is internally consistent; it does not establish query-answering fidelity or interoperability, because the competency-question evaluation is directly contradicted by the implementation details given in Section 4, and no OWL artifact or external dataset is provided. The underlying taxonomy may well be salvageable, but the manuscript as written does not demonstrate its central assertion.","major_comments":[{"comment":"The text states: 'To address CQ1, we associated the data properties LayOffFromJob and CrowdingAtHome with the class subjectOfCare.' Neither property represents residence in a low-income area, so CQ1 ('patients lives in a low-income area') is not answered by the described implementation. This is load-bearing because CQ1 and CQ2 are the only query-answering evidence for the paper's central claim. The authors must either add and query a property that actually represents low-income area, or restate CQ1 to match the implemented properties.","section":"§4, CQ1"},{"comment":"The displayed SPARQL query for CQ2 places CrowdingAtHome inside an OPTIONAL clause without any FILTER. Consequently, a laid-off patient with no crowding data is returned, which contradicts the conjunctive reading of CQ2 ('who are laid off from their job and whose houses are also crowded'). The query should require the crowding property and filter its value to true if the query is to reflect the stated competency question.","section":"§4, CQ2 SPARQL"},{"comment":"No OWL artifact, ontology file, or query artifact is provided, so the DL query in Figure 3 and the claimed HermiT consistency result cannot be independently reproduced or checked. Because the entire evaluation rests on internal query-answering behavior, the missing artifact is a central reproducibility gap. A persistent URL to the OWL file, the exact DL query, and the version of Protégé/HermiT should be supplied.","section":"§3.2, §4"},{"comment":"The paper's interoperability claim is stated as an aim ('CSSDH aims to achieve interoperability') and the evaluation section explicitly defers real data testing to future work ('Our future plan is to test our model with a large dataset from NHS Scotland and HSE Ireland, MIMIC-III ...'). No external EHR data, integration scenario, or cross-system query is used. The title and abstract claim 'Operational Continuity of Care Data Interoperability,' but the evidence supports only internal consistency; the interoperability claim needs either a data-level demonstration or an explicit reframing as a design goal.","section":"§4"}],"minor_comments":[{"comment":"The acronym is inconsistent: the paper uses CSSDH in the title, abstract, and Section 3.1, but CSSDM in Sections 3.2 and 4. Please use one consistent name throughout.","section":"§3.2, §4"},{"comment":"The phrase 'obtained no pitfall badge' is unclear; oops! reports pitfalls rather than badges. Please state the actual number and types of pitfalls detected, or report explicitly that zero pitfalls were found.","section":"§4"},{"comment":"Several cited resources are missing bibliographic entries: ISO/DIS 13940:2024 ContSys, ICD-11 chapter 24, the Gravity project SDH value set, and the Healthy People 2030 initiative. Please add references or persistent URLs for these.","section":"References"},{"comment":"The SPARQL listing is difficult to read because tokens are broken across lines (e.g., 'La y o f f f ro m jo b'). Provide a clean, copyable query listing, preferably in an appendix or auxiliary file.","section":"§4, SPARQL listing"},{"comment":"The Boolean-only encoding for SDH data is justified by privacy considerations, but the paper should also discuss the resulting loss of information granularity, for example how 'crowded' or 'low-income' is defined and whether a binary flag is clinically meaningful for the stated competency questions.","section":"§3.2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript has a genuine gap between its claims and its evidence, but the core idea—an SDH ontology built on ContSys and WHO SDH—is plausible and could be made publishable with a corrected evaluation and a released artifact. I found no indication of misconduct; the main issues are internal inconsistency in the competency-question mapping, the missing OWL artifact, and the unsupported interoperability framing. Given the journal's scope, I would invite a revision rather than reject, provided the authors address the query-level mismatch and supply the artifact."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this is a modest, plausible ontology-engineering contribution that overclaims what its evaluation shows. The authors extend their prior ContSys ontology with SDH concepts from SOHO, ICD-11 and the Gravity Project, and report an OWL model with 171 classes, 141 object properties, and 210 data properties. That is a legitimate extension of an established line of work, and the motivation — capturing social determinants in continuity of care — is real.\n\nWhat the paper does well: it names the right gap, reuses existing standards rather than inventing terms, and runs two standard sanity checks, HermiT consistency and the oops! pitfall scanner, both of which pass. The DL query in Figure 3 may indeed be answerable in the model as built.\n\nNow the soft spots, in order of importance. First, the evaluation does not actually answer its own competency questions. CQ1 asks for patients in a low-income area, but the text says the authors addressed it by associating LayOffFromJob and CrowdingAtHome with subjectOfCare. Those properties have nothing to do with income. Second, the SPARQL query shown for CQ2 uses OPTIONAL for CrowdingAtHome with no FILTER, so it returns laid-off patients even if crowding is unknown. The conjunctive intent of CQ2 is not enforced. Third, there is no OWL file, so the DL query and consistency check cannot be reproduced by a reader. Fourth, the competency questions were authored by the same team to match properties they added, so the evaluation is self-referential. There is no external dataset, no comparison with other SDH models, and no demonstration of interoperability with any real EHR system. The paper also has copy-paste inconsistencies, most visibly CSSDH versus CSSDM.\n\nNone of these flaws is fatal to the underlying idea. A corrected implementation, a released artifact, and at least one external dataset would make the interoperability claim testable. As published, the evidence supports only 'we built a coherent ontology extension,' not 'this enables continuity-of-care data interoperability.'\n\nWho should read it: people building SDH-related ontologies or information models for EHRs will find the class and property choices useful as a starting point. I would not cite it in its current form because the artifact is unavailable and the query results are not reproducible.\n\nRecommendation for peer review: yes, send it, but with a clear expectation of major revision — fix the evaluation, release the OWL file, and either correct the competency-question mapping or state the questions honestly. The manuscript is not ready as is, but the topic is important and the authors appear to be doing real work.","headline":"A plausible ontology extension whose own evaluation fails to support its interoperability claim; fix the evidence and release the artifact.","tokens_in":5220,"tokens_out":1882,"would_cite":false,"duration_ms":17422,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"CSSDH combines ISO/DIS 13940 ContSys with WHO social determinants of health to give continuity-of-care records a formal way to capture and query patients' social context.","keywords":["social determinants of health","continuity of care","ontology","OWL","EHR interoperability","ISO 13940 ContSys","competency questions","privacy"],"falsifier":"Encode a realistic set of patient records with known social determinants in CSSDH, run CQ1 and CQ2 plus independent clinical queries written by someone other than the authors, and attempt to exchange the data with a standards-based EHR system; if any expected patient is missed, any unexpected patient appears, or the social-determinant values are lost in exchange, the interoperability claim fails.","tokens_in":4271,"feed_emoji":"🏥","tokens_out":9957,"duration_ms":85039,"temperature":0.7,"pith_summary":"CSSDH, the Common Semantic Data Model for Social Determinants of Health, is a formal OWL ontology that combines ISO/DIS 13940:2024 ContSys with the WHO's social determinants of health (SDH) concepts. The paper's central claim is that this integration gives electronic health records a structured way to represent and query social factors such as employment, housing, and living conditions across care settings. If true, care teams and policymakers could ask questions like 'which patients are laid off and live in crowded homes?' directly against EHR data. The evidence offered is an internal consistency check, two author-written competency questions, a SPARQL sketch, and a clean pitfall scan.","feed_headline":"CSSDH: a semantic model for social determinants in continuity of care","feed_subtitle":"Gives electronic health records a standard way to capture and query social factors that shape health outcomes.","key_machinery":"The central object is the CSSDH ontology itself, defined in OWL and built in Protégé. Its load-bearing mechanism is attaching SDH-related data properties directly to the subject-of-care class while anchoring clinical structure in ISO/DIS 13940 ContSys and aligning upper-level classes with DOLCE. The argument also depends on the competency-question method (CQ1 and CQ2) as the evaluation instrument, and on the HermiT reasoner and DL queries to test consistency and retrieval. Together these form a demonstration that SDH can be modeled inside a continuity-of-care ontology in a privacy-conscious, queryable way.","core_discovery":"Working in the authors' own terms, the discovery is an integrated formal information model: CSSDH. It reuses terminology from the SOHO ontology, the Social Prescribing Ontology, ICD-11 chapter 24, and the Gravity project value set, organizes it with OntoClean, and attaches social-determinant details to the subject-of-care class as boolean data properties (e.g., LayOffFromJob, CrowdingAtHome). This design is meant to preserve patient privacy while still enabling semantic reasoning and retrieval. The evaluation shows the ontology is consistent under the HermiT reasoner, returns answers for the two competency questions via a DL query and a SPARQL sketch, and raises no pitfalls in the oops! scanner. The intended consequence is that CSSDH operationalizes SDH capture in the continuity-of-care network, closing a gap left by existing SDH ontologies that are term lists rather than formal information models.","pith_inferences":["The paper's own evaluation does not yet establish interoperability with real heterogeneous systems; a stronger test would map CSSDH instances to HL7 FHIR or openEHR and verify that SDH data round-trips between systems.","The boolean-only representation of SDH, chosen for privacy, discards detail such as income brackets or crowding severity; a testable extension is to compare CSSDH's query expressiveness against value-set-based approaches on realistic screening instruments.","Competency questions written by the authors may not reflect real clinical information needs; a robust evaluation would derive questions from standardized SDH screening tools (e.g., the Gravity project's own questionnaires) and measure precision and recall over annotated records.","If the planned large-scale validation succeeds, CSSDH could serve as a reference model for EHR vendors aiming to meet Healthy People 2030 SDH-capture goals while respecting patient privacy."],"forward_implications":["If CSSDH is correct, EHR systems can add social determinants to their data model as boolean patient attributes, making queries about living conditions and employment routine.","The ontology's reuse of ISO 13940 and WHO SDH concepts means it can be aligned with existing continuity-of-care models and extended with other standard vocabularies.","Because social factors are stored as true/false flags, patient privacy is protected while still allowing semantic reasoning over those flags.","The model is positioned to be validated on large clinical datasets (NHS Scotland, HSE Ireland, MIMIC-III, Recoletas Red Hospital), which would test whether it performs outside the two competency questions.","A clean oops! scan and HermiT consistency suggest the ontology is at least internally well-formed, a necessary condition for any semantic interoperability claim."],"supporting_citations":[{"why":"It supplies the WHO operational framework that motivates capturing SDH and the privacy rationale for boolean-valued data.","marker":"[12]"},{"why":"It is the SOHO ontology, the main source of SDH terminology that CSSDH reuses and reorganizes.","marker":"[10]"},{"why":"It is the Social Prescribing Ontology, another terminology source for SDH concepts in CSSDH.","marker":"[9]"},{"why":"It supplies the OntoClean methodology that CSSDH uses to organize and validate its class hierarchy.","marker":"[8]"},{"why":"It provides DOLCE, the top-level ontology on which CSSDH's ContSys alignment builds.","marker":"[6]"},{"why":"It is the authors' previous HL7-FHIR-based ContSys ontology that CSSDH extends with SDH concepts.","marker":"[4]"},{"why":"It provides the competency-question methodology used to evaluate CSSDH via CQ1 and CQ2.","marker":"[7]"},{"why":"It supports the claim that existing SDH ontologies are term lists rather than formal information models, motivating CSSDH.","marker":"[5]"}],"fun_headline_variants":["CSSDH ontology: standardizing social determinants in care","New CSSDH model ties social determinants to care data interoperability","Ontology CSSDH makes social health factors queryable in EHRs","CSSDH: semantic ontology for social determinants in continuity of care"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that passing a HermiT consistency check, one DL query, one SPARQL sketch, and an oops! scan against two author-written competency questions proves the model is logically sound and will support real EHR interoperability.","fun_headline_variants_meta":{"raw":{"variants":["CSSDH ontology: standardizing social determinants in care","New CSSDH model ties social determinants to care data interoperability","Ontology CSSDH makes social health factors queryable in EHRs","CSSDH: semantic ontology for social determinants in continuity of care"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00022,"raw_usage":{"total_tokens":1440,"prompt_tokens":931,"completion_tokens":509,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":547,"completion_tokens_details":{"reasoning_tokens":440}},"tokens_in":547,"tokens_out":509,"duration_ms":5803,"temperature":1.0,"reasoning_tokens":440,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T17:11:31.489233+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Encode a realistic set of patient records with known social determinants in CSSDH, run CQ1 and CQ2 plus independent clinical queries written by someone other than the authors, and attempt to exchange the data with a standards-based EHR system; if any expected patient is missed, any unexpected patient appears, or the social-determinant values are lost in exchange, the interoperability claim fails.","supporting_citations":[{"cited_title":"World Health Organization (2024)","cited_arxiv_id":null,"evidence_quote":"It supplies the WHO operational framework that motivates capturing SDH and the privacy rationale for boolean-valued data."},{"cited_title":"In: 2022 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)","cited_arxiv_id":null,"evidence_quote":"It is the SOHO ontology, the main source of SDH terminology that CSSDH reuses and reorganizes."},{"cited_title":"Journal of Medical Internet Research 22(12), e23721 (2020)","cited_arxiv_id":null,"evidence_quote":"It is the Social Prescribing Ontology, another terminology source for SDH concepts in CSSDH."},{"cited_title":"In: International conference on knowledge engineering and knowledge management","cited_arxiv_id":null,"evidence_quote":"It provides DOLCE, the top-level ontology on which CSSDH's ContSys alignment builds."},{"cited_title":"Journal of Personalized Medicine 13(7), 1024 (2023)","cited_arxiv_id":null,"evidence_quote":"It is the authors' previous HL7-FHIR-based ContSys ontology that CSSDH extends with SDH concepts."},{"cited_title":"In: Benchmarking —Theory and practice, pp","cited_arxiv_id":null,"evidence_quote":"It provides the competency-question methodology used to evaluate CSSDH via CQ1 and CQ2."},{"cited_title":"IEEE Access 7, 91265 –91290 (2019)","cited_arxiv_id":null,"evidence_quote":"It supports the claim that existing SDH ontologies are term lists rather than formal information models, motivating CSSDH."}],"review_version":1}