Disease Classifications

Classification of channelopathies by primary organ system affected. Used to tag diseases caused by ion channel dysfunction.

The regional legal schedule of occupational disease for the European Union: Annex I and Annex II of Commission Recommendation 2003/670/EC of 19 September 2003 concerning the European schedule of occupational diseases, **as amended**. This enum encodes the schedule in its **current consolidated form**, which is the version established by Commission Recommendation (EU) 2025/2609 of 18 December 2025. Two amendments have changed the schedule since 2003: - **Recommendation (EU) 2022/2337** (28 November 2022) added item **408**, COVID-19 contracted through work in disease prevention, health and social care and domiciliary assistance, or in a pandemic context in sectors with a proven outbreak risk. - **Recommendation (EU) 2025/2609** (18 December 2025) added four asbestos-related diseases to Annex I — **311** cancer of the larynx, **312** cancer of the ovary, **313** pleural plaques with functional impairment, and **314** non-malignant pleural effusion — and three suspected asbestos-related cancers to Annex II (**2.309** colon, **2.310** rectum, **2.311** stomach). In the same revision Annex II item 2.308 (cancer of the larynx following the inhalation of asbestos dust) was **removed**, having been promoted into Annex I as item 311; it is therefore not a permissible value here. Member States were asked to report implementation by 31 December 2026. Relationship to the ILO list. This schedule and the ILO list (``ILOCausativeAgentEnum`` / ``ILODiseaseCategoryEnum``) are separate instruments over the same subject matter and are **not** substitutes: the ILO list is the global tripartite consensus and is *biaxial*, organising section 2 by target organ system, which is why it takes two enums and two slots here. The European schedule is a regional instrument organised throughout by causative agent or route, so it is single-axis and takes one enum and one slot; it joins the ILO slots in the presentational ``occupational_classification`` slot group. The European schedule is also more specific in places the ILO list is generic — it names silicosis, asbestosis and mesothelioma as distinct items (301.11, 301.21, 301.22) where the ILO list has one fibrogenic-mineral-dust pneumoconiosis item (2.1.1) — and it uniquely carries COVID-19 and the asbestos additions above. Assign both when both apply; neither implies the other. Two tiers, and they mean different things: - **Annex I** is the recognised schedule. Member States are asked to introduce these diseases into national law; a listed disease linked directly to the occupation is liable for compensation and subject to prevention measures. - **Annex II** is the *additional list of diseases suspected of being occupational in origin* — notifiable, and candidates for later inclusion in Annex I. An Annex II assignment records a suspected, not established, occupational aetiology. Annex II keys are prefixed ``suspected_`` so the two tiers can never be confused at a glance, and an Annex II assignment must say so in ``notes``. Structure. Both annexes use the same five chapters (chemical agents; skin diseases; diseases caused by inhalation; infectious and parasitic diseases; physical agents), and items are numbered — Annex I ``100``-``508``, Annex II ``2.101``-``2.503``. Chapters 1 and 2.1 name *agents* rather than diseases, so an item there covers whatever disease that agent causes at work. Items 201 and 301 are themselves headers with numbered sub-items, and are encoded as intermediate nodes. The published numbering has genuine gaps (there is no 115.03, 134, 304.03, 305.02, 501, 2.131-2.139, 2.302 or 2.306) and, in two places, sub-items are printed out of numeric order (201.01/201.03/201.02/201.04, and 304.05/305.01/ 304.06). Both are faithfully reproduced: the gaps are simply absent, and item order follows the published text, not numeric sort. Do not "repair" either. Curation guidance for dismech: - Assign only when the disease is recognised as occupational in origin, on the same reasoning as the ILO enum — the schedule requires that the disease "must be linked directly to the occupation". - Record provenance in ``notes``: the item number, the annex, and the amendment that introduced it where it is one of the post-2003 additions. As with ICIMD and ISDS this is a definitional taxonomy mapping, not an empirical disease claim, so prefer ``notes`` over a manufactured evidence ``snippet``. - Item numbers are stable within the consolidated schedule and are what national law cites, so each is recorded verbatim at the head of its ``description``. Following the repo convention set by ``ISDSNosologyGroupEnum``, numbers are not part of any permissible-value key, so a future renumbering does not invalidate stored assignments. Sources: Commission Recommendation 2003/670/EC (OJ L 238, 25.9.2003, p. 28); Commission Recommendation (EU) 2022/2337 (OJ L 309, 30.11.2022); Commission Recommendation (EU) 2025/2609 (OJ L, 2025/2609). http://data.europa.eu/eli/reco/2003/670/oj

Sanctioned classification axes for an **exposure or its agent** — as distinct from the disease-level nosologies in ``DiseaseClassifications``. This is the load-bearing distinction in this module. ``ILOCausativeAgentEnum``, ``ILODiseaseCategoryEnum`` and ``EUOccupationalScheduleEnum`` classify a *disease*, so they hang off ``Disease.classifications`` — note that even the ILO's causative-agent axis is a statement about which agent-indexed item a *disease* is recognised under, not a property of the agent itself. Everything here classifies the *agent or the exposure event* — "benzene is an IARC Group 1 carcinogen" is a statement about benzene, not about any disease benzene causes. Putting agent-level facts into a disease-level slot would assert, wrongly, that the disease itself carries the hazard classification. These enums are therefore reached through ``Environmental.exposure_classifications``. Each axis below is an independent, published standard; they are deliberately separate enums rather than one blended vocabulary, because an exposure has a value on several of them at once (chronic + inhalation + chemical + IARC Group 1 is a normal, non-redundant description of occupational benzene). These annotate an exposure that dismech has already grounded to ECTO/ExO via ``Environmental.exposure_term``. They do not replace that grounding — ECTO post-composes the exposure event itself ("exposure to arsenic in water via ingestion"), while these axes record the regulatory and toxicological classification of the agent and the exposure's shape. Where an ECTO term already encodes the route, the ``exposure_route`` value simply makes it queryable without parsing the term label. Scope note — carcinogen classifications. Only the IARC Monographs classification is encoded here. Three national systems classify the same agents on parallel scales — the US NTP *Report on Carcinogens* ("known to be a human carcinogen" / "reasonably anticipated to be a human carcinogen"), the US EPA IRIS cancer descriptors (five weight-of-evidence descriptors), and the ACGIH TLV carcinogenicity notations (A1-A5). They are deliberately **not** encoded: they largely re-rank the same evidence, and carrying four near-parallel scales invites curators to assert an NTP or ACGIH listing they have not checked. IARC is the WHO/international instrument and is the one cited in the occupational literature dismech curates. If a national listing is the specific point being made, state it in the assignment's ``notes`` with its source. Also deliberately not encoded: the GBD (IHME) environmental and occupational risk-factor hierarchy. It is a genuine sanctioned classification and would fit this module, but its exact Level 2 / Level 3 label set could not be verified against a primary source when this module was written, and transcribing an approximate hierarchy would defeat the purpose. Left for a follow-up that works from the GBD capstone appendix.

Sanctioned classification axes for an **exposure or its agent** — as distinct from the disease-level nosologies in ``DiseaseClassifications``. This is the load-bearing distinction in this module. ``ILOCausativeAgentEnum``, ``ILODiseaseCategoryEnum`` and ``EUOccupationalScheduleEnum`` classify a *disease*, so they hang off ``Disease.classifications`` — note that even the ILO's causative-agent axis is a statement about which agent-indexed item a *disease* is recognised under, not a property of the agent itself. Everything here classifies the *agent or the exposure event* — "benzene is an IARC Group 1 carcinogen" is a statement about benzene, not about any disease benzene causes. Putting agent-level facts into a disease-level slot would assert, wrongly, that the disease itself carries the hazard classification. These enums are therefore reached through ``Environmental.exposure_classifications``. Each axis below is an independent, published standard; they are deliberately separate enums rather than one blended vocabulary, because an exposure has a value on several of them at once (chronic + inhalation + chemical + IARC Group 1 is a normal, non-redundant description of occupational benzene). These annotate an exposure that dismech has already grounded to ECTO/ExO via ``Environmental.exposure_term``. They do not replace that grounding — ECTO post-composes the exposure event itself ("exposure to arsenic in water via ingestion"), while these axes record the regulatory and toxicological classification of the agent and the exposure's shape. Where an ECTO term already encodes the route, the ``exposure_route`` value simply makes it queryable without parsing the term label. Scope note — carcinogen classifications. Only the IARC Monographs classification is encoded here. Three national systems classify the same agents on parallel scales — the US NTP *Report on Carcinogens* ("known to be a human carcinogen" / "reasonably anticipated to be a human carcinogen"), the US EPA IRIS cancer descriptors (five weight-of-evidence descriptors), and the ACGIH TLV carcinogenicity notations (A1-A5). They are deliberately **not** encoded: they largely re-rank the same evidence, and carrying four near-parallel scales invites curators to assert an NTP or ACGIH listing they have not checked. IARC is the WHO/international instrument and is the one cited in the occupational literature dismech curates. If a national listing is the specific point being made, state it in the assignment's ``notes`` with its source. Also deliberately not encoded: the GBD (IHME) environmental and occupational risk-factor hierarchy. It is a genuine sanctioned classification and would fit this module, but its exact Level 2 / Level 3 label set could not be verified against a primary source when this module was written, and transcribing an approximate hierarchy would defeat the purpose. Left for a follow-up that works from the GBD capstone appendix.

Sanctioned classification axes for an **exposure or its agent** — as distinct from the disease-level nosologies in ``DiseaseClassifications``. This is the load-bearing distinction in this module. ``ILOCausativeAgentEnum``, ``ILODiseaseCategoryEnum`` and ``EUOccupationalScheduleEnum`` classify a *disease*, so they hang off ``Disease.classifications`` — note that even the ILO's causative-agent axis is a statement about which agent-indexed item a *disease* is recognised under, not a property of the agent itself. Everything here classifies the *agent or the exposure event* — "benzene is an IARC Group 1 carcinogen" is a statement about benzene, not about any disease benzene causes. Putting agent-level facts into a disease-level slot would assert, wrongly, that the disease itself carries the hazard classification. These enums are therefore reached through ``Environmental.exposure_classifications``. Each axis below is an independent, published standard; they are deliberately separate enums rather than one blended vocabulary, because an exposure has a value on several of them at once (chronic + inhalation + chemical + IARC Group 1 is a normal, non-redundant description of occupational benzene). These annotate an exposure that dismech has already grounded to ECTO/ExO via ``Environmental.exposure_term``. They do not replace that grounding — ECTO post-composes the exposure event itself ("exposure to arsenic in water via ingestion"), while these axes record the regulatory and toxicological classification of the agent and the exposure's shape. Where an ECTO term already encodes the route, the ``exposure_route`` value simply makes it queryable without parsing the term label. Scope note — carcinogen classifications. Only the IARC Monographs classification is encoded here. Three national systems classify the same agents on parallel scales — the US NTP *Report on Carcinogens* ("known to be a human carcinogen" / "reasonably anticipated to be a human carcinogen"), the US EPA IRIS cancer descriptors (five weight-of-evidence descriptors), and the ACGIH TLV carcinogenicity notations (A1-A5). They are deliberately **not** encoded: they largely re-rank the same evidence, and carrying four near-parallel scales invites curators to assert an NTP or ACGIH listing they have not checked. IARC is the WHO/international instrument and is the one cited in the occupational literature dismech curates. If a national listing is the specific point being made, state it in the assignment's ``notes`` with its source. Also deliberately not encoded: the GBD (IHME) environmental and occupational risk-factor hierarchy. It is a genuine sanctioned classification and would fit this module, but its exact Level 2 / Level 3 label set could not be verified against a primary source when this module was written, and transcribing an approximate hierarchy would defeat the purpose. Left for a follow-up that works from the GBD capstone appendix.

Sanctioned classification axes for an **exposure or its agent** — as distinct from the disease-level nosologies in ``DiseaseClassifications``. This is the load-bearing distinction in this module. ``ILOCausativeAgentEnum``, ``ILODiseaseCategoryEnum`` and ``EUOccupationalScheduleEnum`` classify a *disease*, so they hang off ``Disease.classifications`` — note that even the ILO's causative-agent axis is a statement about which agent-indexed item a *disease* is recognised under, not a property of the agent itself. Everything here classifies the *agent or the exposure event* — "benzene is an IARC Group 1 carcinogen" is a statement about benzene, not about any disease benzene causes. Putting agent-level facts into a disease-level slot would assert, wrongly, that the disease itself carries the hazard classification. These enums are therefore reached through ``Environmental.exposure_classifications``. Each axis below is an independent, published standard; they are deliberately separate enums rather than one blended vocabulary, because an exposure has a value on several of them at once (chronic + inhalation + chemical + IARC Group 1 is a normal, non-redundant description of occupational benzene). These annotate an exposure that dismech has already grounded to ECTO/ExO via ``Environmental.exposure_term``. They do not replace that grounding — ECTO post-composes the exposure event itself ("exposure to arsenic in water via ingestion"), while these axes record the regulatory and toxicological classification of the agent and the exposure's shape. Where an ECTO term already encodes the route, the ``exposure_route`` value simply makes it queryable without parsing the term label. Scope note — carcinogen classifications. Only the IARC Monographs classification is encoded here. Three national systems classify the same agents on parallel scales — the US NTP *Report on Carcinogens* ("known to be a human carcinogen" / "reasonably anticipated to be a human carcinogen"), the US EPA IRIS cancer descriptors (five weight-of-evidence descriptors), and the ACGIH TLV carcinogenicity notations (A1-A5). They are deliberately **not** encoded: they largely re-rank the same evidence, and carrying four near-parallel scales invites curators to assert an NTP or ACGIH listing they have not checked. IARC is the WHO/international instrument and is the one cited in the occupational literature dismech curates. If a national listing is the specific point being made, state it in the assignment's ``notes`` with its source. Also deliberately not encoded: the GBD (IHME) environmental and occupational risk-factor hierarchy. It is a genuine sanctioned classification and would fit this module, but its exact Level 2 / Level 3 label set could not be verified against a primary source when this module was written, and transcribing an approximate hierarchy would defeat the purpose. Left for a follow-up that works from the GBD capstone appendix.

Sanctioned classification axes for an **exposure or its agent** — as distinct from the disease-level nosologies in ``DiseaseClassifications``. This is the load-bearing distinction in this module. ``ILOCausativeAgentEnum``, ``ILODiseaseCategoryEnum`` and ``EUOccupationalScheduleEnum`` classify a *disease*, so they hang off ``Disease.classifications`` — note that even the ILO's causative-agent axis is a statement about which agent-indexed item a *disease* is recognised under, not a property of the agent itself. Everything here classifies the *agent or the exposure event* — "benzene is an IARC Group 1 carcinogen" is a statement about benzene, not about any disease benzene causes. Putting agent-level facts into a disease-level slot would assert, wrongly, that the disease itself carries the hazard classification. These enums are therefore reached through ``Environmental.exposure_classifications``. Each axis below is an independent, published standard; they are deliberately separate enums rather than one blended vocabulary, because an exposure has a value on several of them at once (chronic + inhalation + chemical + IARC Group 1 is a normal, non-redundant description of occupational benzene). These annotate an exposure that dismech has already grounded to ECTO/ExO via ``Environmental.exposure_term``. They do not replace that grounding — ECTO post-composes the exposure event itself ("exposure to arsenic in water via ingestion"), while these axes record the regulatory and toxicological classification of the agent and the exposure's shape. Where an ECTO term already encodes the route, the ``exposure_route`` value simply makes it queryable without parsing the term label. Scope note — carcinogen classifications. Only the IARC Monographs classification is encoded here. Three national systems classify the same agents on parallel scales — the US NTP *Report on Carcinogens* ("known to be a human carcinogen" / "reasonably anticipated to be a human carcinogen"), the US EPA IRIS cancer descriptors (five weight-of-evidence descriptors), and the ACGIH TLV carcinogenicity notations (A1-A5). They are deliberately **not** encoded: they largely re-rank the same evidence, and carrying four near-parallel scales invites curators to assert an NTP or ACGIH listing they have not checked. IARC is the WHO/international instrument and is the one cited in the occupational literature dismech curates. If a national listing is the specific point being made, state it in the assignment's ``notes`` with its source. Also deliberately not encoded: the GBD (IHME) environmental and occupational risk-factor hierarchy. It is a genuine sanctioned classification and would fit this module, but its exact Level 2 / Level 3 label set could not be verified against a primary source when this module was written, and transcribing an approximate hierarchy would defeat the purpose. Left for a follow-up that works from the GBD capstone appendix.

Sanctioned classification axes for an **exposure or its agent** — as distinct from the disease-level nosologies in ``DiseaseClassifications``. This is the load-bearing distinction in this module. ``ILOCausativeAgentEnum``, ``ILODiseaseCategoryEnum`` and ``EUOccupationalScheduleEnum`` classify a *disease*, so they hang off ``Disease.classifications`` — note that even the ILO's causative-agent axis is a statement about which agent-indexed item a *disease* is recognised under, not a property of the agent itself. Everything here classifies the *agent or the exposure event* — "benzene is an IARC Group 1 carcinogen" is a statement about benzene, not about any disease benzene causes. Putting agent-level facts into a disease-level slot would assert, wrongly, that the disease itself carries the hazard classification. These enums are therefore reached through ``Environmental.exposure_classifications``. Each axis below is an independent, published standard; they are deliberately separate enums rather than one blended vocabulary, because an exposure has a value on several of them at once (chronic + inhalation + chemical + IARC Group 1 is a normal, non-redundant description of occupational benzene). These annotate an exposure that dismech has already grounded to ECTO/ExO via ``Environmental.exposure_term``. They do not replace that grounding — ECTO post-composes the exposure event itself ("exposure to arsenic in water via ingestion"), while these axes record the regulatory and toxicological classification of the agent and the exposure's shape. Where an ECTO term already encodes the route, the ``exposure_route`` value simply makes it queryable without parsing the term label. Scope note — carcinogen classifications. Only the IARC Monographs classification is encoded here. Three national systems classify the same agents on parallel scales — the US NTP *Report on Carcinogens* ("known to be a human carcinogen" / "reasonably anticipated to be a human carcinogen"), the US EPA IRIS cancer descriptors (five weight-of-evidence descriptors), and the ACGIH TLV carcinogenicity notations (A1-A5). They are deliberately **not** encoded: they largely re-rank the same evidence, and carrying four near-parallel scales invites curators to assert an NTP or ACGIH listing they have not checked. IARC is the WHO/international instrument and is the one cited in the occupational literature dismech curates. If a national listing is the specific point being made, state it in the assignment's ``notes`` with its source. Also deliberately not encoded: the GBD (IHME) environmental and occupational risk-factor hierarchy. It is a genuine sanctioned classification and would fit this module, but its exact Level 2 / Level 3 label set could not be verified against a primary source when this module was written, and transcribing an approximate hierarchy would defeat the purpose. Left for a follow-up that works from the GBD capstone appendix.

Part-level classification for organizing disorders by organ system and clinical domain, following the high-level structure of Harrison's Principles of Internal Medicine.

Histological/morphological classification of cancers based on ICD-O morphology axis. Applies only to neoplastic diseases. Single-inheritance hierarchy.

The international standard nosology of occupational disease: the **List of occupational diseases (revised 2010)**, annexed to the List of Occupational Diseases Recommendation, 2002 (No. 194) of the International Labour Organization. The 2010 revision was approved by the ILO Governing Body on 25 March 2010 at its 307th Session, on the basis of two tripartite Meetings of Experts (2005 and 2009), and replaced the list adopted in 2002. This is the closest thing that exists to a globally sanctioned nosology of work-related disease. Recommendation No. 194 asks member States to establish a national list of occupational diseases for prevention, recording, notification and compensation, comprising "to the extent possible" the diseases in this annex; national schedules are therefore derived from it, and it is the common denominator across them. It is a *recognition* instrument rather than a mechanistic one — an item asserts that a causal link between a work exposure and a disease is established well enough to be recognised, which is exactly the claim dismech wants to record when it classifies an entry as occupational. **The list is biaxial, and dismech models that as two enums.** The instrument has four sections built on two different organising principles: - Section 1 ("caused by exposure to agents arising from work activities") and section 3.1 ("Cancer caused by the following agents") index by **causative agent** — the item names the agent, e.g. "Diseases caused by benzene or its homologues". - Section 2 ("by target organ systems") and section 4 ("Other diseases") index by **the disease itself** — the item names a clinical entity, e.g. "Asthma caused by recognized sensitizing agents", "Miners' nystagmus". Because the axes are orthogonal, one disease legitimately takes an item from each: occupational asthma from isocyanates is both 1.1.35 (agent) and 2.1.7 (disease). Collapsing both axes into one enum would make it impossible for the schema to say which axis a value belongs to, so the values are split into ``ILOCausativeAgentEnum`` (sections 1 and 3) and ``ILODiseaseCategoryEnum`` (sections 2 and 4), reached by two distinct slots that share a presentational ``slot_group``. This is the opposite of the ISDS nosology's rule, where a disorder is listed once by construction; do not carry the ISDS "assign exactly one" habit over to this instrument. Two consequences of the split a curator should know: - **The ILO's contiguous item numbering is spread across two enums.** Items 1.x and 3.x live in ``ILOCausativeAgentEnum``; items 2.x and 4.x live in ``ILODiseaseCategoryEnum``. Each value records its item number verbatim at the head of its ``description``, so a number lookup still works, but you cannot assume one enum holds the whole list. - **Placing sections 3 and 4 is dismech's reading, not the ILO's.** The instrument presents four sections, not two axes. Section 3.1 is put on the agent axis because its own title indexes by agent; section 4 is put on the disease axis because its one named item is a clinical entity. If that reading is ever revised, the affected values move between enums. Two wrinkles in the instrument itself, reproduced rather than tidied: - **Section 1.3 names diseases, not agents.** Its title, "Biological agents and infectious or parasitic diseases", acknowledges the hybrid, and its items are Brucellosis, Tetanus, Tuberculosis and so on. It stays on the agent axis because the ILO places it under section 1. - **Both slots stay multivalued**, because more than one item from a single axis is normal: silicosis takes both 2.1.1 (pneumoconiosis) and 2.1.2 (silicotuberculosis). Open items. Every subsection ends with an "open item" (1.1.41, 1.2.7, 1.3.9, 2.1.12, 2.2.4, 2.3.8, 2.4.2, 3.1.21, 4.2) permitting recognition of a disease the list does not name, where a direct link to work exposure is established scientifically. These are transcribed as real permissible values because the instrument uses them substantively, but they are a weak assignment: prefer a specific item, and when using an open item say in ``notes`` what the agent or disease actually was. Mental and behavioural disorders (2.4) and the musculoskeletal section were new in the 2010 revision — the first time the ILO list named a mental disorder (post-traumatic stress disorder) explicitly. Curation guidance for dismech: - Assign only when the disease is genuinely recognised as occupational in origin, not merely because an exposure exists. Lead poisoning from a contaminated water supply is not ILO 1.1.8; lead poisoning in a smelter worker is. - A disease that occurs both occupationally and non-occupationally (asthma, COPD, mesothelioma, hearing loss) still takes the item — the assignment records that an occupational form is recognised, not that every case is occupational. Say which in ``notes``. - Record provenance in ``notes``: the revision (2010), the item number, and the item text where it differs from the dismech entry name. As with ICIMD and ISDS, this is a definitional taxonomy mapping, not an empirical disease claim, so ``notes`` is preferred over a manufactured evidence ``snippet``. - Pair with ``harrisons_chapter`` — usually ``ENVIRONMENTAL_EXPOSURES``, ``POISONING_ENVENOMATION``, or the relevant organ-system Part. Item numbers here ARE stable within the 2010 revision and are what national law and statistical reporting cite. They are NOT stable across revisions (the 2002 list numbered differently), which is why, following the repo convention established by ``ISDSNosologyGroupEnum``, numbers are not part of any permissible-value key.

The international standard nosology of occupational disease: the **List of occupational diseases (revised 2010)**, annexed to the List of Occupational Diseases Recommendation, 2002 (No. 194) of the International Labour Organization. The 2010 revision was approved by the ILO Governing Body on 25 March 2010 at its 307th Session, on the basis of two tripartite Meetings of Experts (2005 and 2009), and replaced the list adopted in 2002. This is the closest thing that exists to a globally sanctioned nosology of work-related disease. Recommendation No. 194 asks member States to establish a national list of occupational diseases for prevention, recording, notification and compensation, comprising "to the extent possible" the diseases in this annex; national schedules are therefore derived from it, and it is the common denominator across them. It is a *recognition* instrument rather than a mechanistic one — an item asserts that a causal link between a work exposure and a disease is established well enough to be recognised, which is exactly the claim dismech wants to record when it classifies an entry as occupational. **The list is biaxial, and dismech models that as two enums.** The instrument has four sections built on two different organising principles: - Section 1 ("caused by exposure to agents arising from work activities") and section 3.1 ("Cancer caused by the following agents") index by **causative agent** — the item names the agent, e.g. "Diseases caused by benzene or its homologues". - Section 2 ("by target organ systems") and section 4 ("Other diseases") index by **the disease itself** — the item names a clinical entity, e.g. "Asthma caused by recognized sensitizing agents", "Miners' nystagmus". Because the axes are orthogonal, one disease legitimately takes an item from each: occupational asthma from isocyanates is both 1.1.35 (agent) and 2.1.7 (disease). Collapsing both axes into one enum would make it impossible for the schema to say which axis a value belongs to, so the values are split into ``ILOCausativeAgentEnum`` (sections 1 and 3) and ``ILODiseaseCategoryEnum`` (sections 2 and 4), reached by two distinct slots that share a presentational ``slot_group``. This is the opposite of the ISDS nosology's rule, where a disorder is listed once by construction; do not carry the ISDS "assign exactly one" habit over to this instrument. Two consequences of the split a curator should know: - **The ILO's contiguous item numbering is spread across two enums.** Items 1.x and 3.x live in ``ILOCausativeAgentEnum``; items 2.x and 4.x live in ``ILODiseaseCategoryEnum``. Each value records its item number verbatim at the head of its ``description``, so a number lookup still works, but you cannot assume one enum holds the whole list. - **Placing sections 3 and 4 is dismech's reading, not the ILO's.** The instrument presents four sections, not two axes. Section 3.1 is put on the agent axis because its own title indexes by agent; section 4 is put on the disease axis because its one named item is a clinical entity. If that reading is ever revised, the affected values move between enums. Two wrinkles in the instrument itself, reproduced rather than tidied: - **Section 1.3 names diseases, not agents.** Its title, "Biological agents and infectious or parasitic diseases", acknowledges the hybrid, and its items are Brucellosis, Tetanus, Tuberculosis and so on. It stays on the agent axis because the ILO places it under section 1. - **Both slots stay multivalued**, because more than one item from a single axis is normal: silicosis takes both 2.1.1 (pneumoconiosis) and 2.1.2 (silicotuberculosis). Open items. Every subsection ends with an "open item" (1.1.41, 1.2.7, 1.3.9, 2.1.12, 2.2.4, 2.3.8, 2.4.2, 3.1.21, 4.2) permitting recognition of a disease the list does not name, where a direct link to work exposure is established scientifically. These are transcribed as real permissible values because the instrument uses them substantively, but they are a weak assignment: prefer a specific item, and when using an open item say in ``notes`` what the agent or disease actually was. Mental and behavioural disorders (2.4) and the musculoskeletal section were new in the 2010 revision — the first time the ILO list named a mental disorder (post-traumatic stress disorder) explicitly. Curation guidance for dismech: - Assign only when the disease is genuinely recognised as occupational in origin, not merely because an exposure exists. Lead poisoning from a contaminated water supply is not ILO 1.1.8; lead poisoning in a smelter worker is. - A disease that occurs both occupationally and non-occupationally (asthma, COPD, mesothelioma, hearing loss) still takes the item — the assignment records that an occupational form is recognised, not that every case is occupational. Say which in ``notes``. - Record provenance in ``notes``: the revision (2010), the item number, and the item text where it differs from the dismech entry name. As with ICIMD and ISDS, this is a definitional taxonomy mapping, not an empirical disease claim, so ``notes`` is preferred over a manufactured evidence ``snippet``. - Pair with ``harrisons_chapter`` — usually ``ENVIRONMENTAL_EXPOSURES``, ``POISONING_ENVENOMATION``, or the relevant organ-system Part. Item numbers here ARE stable within the 2010 revision and are what national law and statistical reporting cite. They are NOT stable across revisions (the 2002 list numbered differently), which is why, following the repo convention established by ``ISDSNosologyGroupEnum``, numbers are not part of any permissible-value key.

Mechanistic (biochemical-pathway) classification of inherited metabolic disorders following the consensus International Classification of Inherited Metabolic Disorders (ICIMD; Ferreira CR, Rahman S, Keller M, Zschocke J & ICIMD Advisory Group, J Inherit Metab Dis 2021;44(1):164-177; PMID:33340416, DOI:10.1002/jimd.12348). ICIMD is a three-level hierarchy — 24 categories (layer 1) over 124 disease groups (layer 2) over ~1,450 individual disorders (layer 3). This enum encodes layers 1 and 2 as a single hierarchical permissible-value set: each category is a top-level value, and each group declares its parent category via ``is_a``. Curators should assign the most specific applicable node (usually a group); the parent category is then derivable through ``is_a``. Both category- and group-level values are valid assignments, so an entry may be tagged at category level when the specific group is unknown. Provenance note: the category set and group membership are transcribed from the ICIMD paper's Results section (the 24 numbered categories) and its Figure 1 sunburst (the group ring). The groups here are the primary ("group"-ring) nodes of that sunburst; a handful of the deepest glycosylation and glycerophospholipid subdivisions are folded into their parent group, so this enum carries 113 groups rather than the paper's headline 124. The eight ICIMD "super-domains" (e.g. "Intermediary metabolism: nutrients") sit ABOVE the category level and are recorded in the category descriptions rather than as separate enum nodes.

Expert-consensus nosology of the genetic skeletal disorders, following the Nosology of Genetic Skeletal Disorders maintained by the Nosology Committee of the International Skeletal Dysplasia Society (ISDS). This enum encodes the **2023 revision** — the eleventh edition — which contains 771 entries associated with 552 genes, classified into 41 groups (Unger S, Ferreira CR, Mortier GR, et al., Am J Med Genet A 2023;191(5):1164-1209; PMID:36779427, DOI:10.1002/ajmg.a.63132). It supersedes the 2019 revision (10th edition; 461 disorders, 437 genes, 42 groups; Mortier GR et al., PMID:31633310), which dismech encoded first and which is still recorded here as provenance — see "Revision handling" below. The ISDS groups are a flat, non-hierarchical partition: each disorder is deliberately listed exactly once, to avoid redundancy in the Nosology. Group membership is therefore mutually exclusive within the nosology itself, even though the underlying biology frequently spans groups (the table carries explicit "see also" cross-references for those cases). Groups mix organizing principles by design — some molecular (a shared causal gene or gene family), some radiographic (which segment of the growing bone is affected), some anatomical or pathogenetic (craniosynostosis, brachydactyly, osteolysis). Dyadic naming. The headline change in the 2023 revision is the adoption of **dyadic naming**: a phenotypic entity is systematically paired with the gene it arises from ("Geleophysic dysplasia, ADAMTSL2-related"), replacing list numberings and eponyms, which the committee considers "more informative and less prone to errors". That is a claim about disease-entity naming and identity, not about this classification axis, and it is deliberately NOT imported into dismech entry naming here; it is noted so curators reading the 2023 table are not surprised by the disorder names. Curation guidance for dismech: - Assign a group only to entries the ISDS Nosology itself lists, or to entries that are an unambiguous subtype/synonym of a listed disorder. This is a transcription of an expert nosology, not an inference engine; do not extend it to skeletal-phenotype disorders the committee chose not to list. - The slot is multivalued only to accommodate an entry that lumps several distinct nosology disorders. A single listed disorder should carry exactly one group. - Record the provenance in the assignment's ``notes`` — which revision, which group, and the listed disorder name where it differs from the dismech entry name. Neither paper's PubMed record carries per-disorder group placements in its abstract, so a group assignment is not quotable as an evidence snippet; prefer ``notes`` over a snippet the abstract does not actually support. - Do NOT use a deprecated value for new curation. Four values are retained only so existing assignments stay resolvable. Revision handling. Permissible-value **keys are stable identifiers and are not renamed or renumbered when a revision renames or renumbers a group** — the key identifies the group across editions, while the revision's own number and name live in the ``description``. Superseded names are retained as ``structured_aliases`` with ``predicate: EXACT_SYNONYM`` and ``source`` pointing at the revision that used them, so a search for the old name still resolves. Group numbers are emphatically NOT stable across revisions (the brachydactyly groups went from 37/38 to 18/19), which is why they are not part of any key. Where a revision *dissolves* a group rather than renaming it, the old value is kept with ``deprecated:`` plus ``deprecated_element_has_possible_replacement`` pointing at its successor — "possible" rather than "exact" because a merge makes the successor broader than the value it replaces. Four 2019 groups are deprecated on that basis: the Perlecan and Aggrecan groups (merged into Proteoglycan core proteins disorders) and the Neonatal osteosclerotic dysplasias and Other sclerosing bone disorders groups (fused into Osteosclerotic disorders). Known gap: the exemplar disorders named in each description below were transcribed from Table 1 of the **2019** revision and have been renumbered and corrected only where the 2023 paper explicitly says a disorder moved (e.g. trichorhinophalangeal dysplasia out of the acromelic group). They are illustrative, not the full membership of a group, and a full re-transcription against the 2023 table (774 rows, extracted and available) is outstanding. The same caveat applies to the per-entry assignments: the bulk of them were derived from the 2019 table and carry 2019 provenance in their ``notes``. That re-verification is tracked in monarch-initiative/dismech#7867, which also records the groups with no dismech coverage at all - including the new 2023 group 28 (parathyroid hormone signaling cascade), whose six disorders (Jansen and Csukasi-Krakow metaphyseal dysplasia, Blomstrand dysplasia, Eiken dysplasia, PTHLH brachydactyly and osteolysis) are simply not yet curated. MONDO mapping policy. No group carries a ``meaning:`` — that would assert the value *is* an ontology class, which is never quite true for a curated nosology group. Where a MONDO class denotes the same disease family it is recorded as ``close_mappings:`` instead, under a deliberately high bar: a candidate is rejected if the MONDO class contains any entity the Nosology itself lists in a *different* group, since such a mapping would silently contradict the committee's own placement. Three of the groups qualify, all gene-defined series that nest cleanly inside one group: - FGFR3 chondrodysplasias → MONDO:0019685 FGFR3-related chondrodysplasia. Contains exactly the group's members and, importantly, excludes the FGFR3 craniosynostoses (Muenke, Crouzon with acanthosis nigricans), which the nosology places in the craniosynostosis group. - TRPV4 disorders → MONDO:0018240 TRPV4-related bone disorder. - Acromesomelic dysplasias → MONDO:0019696 acromesomelic dysplasia. Rejected candidates, recorded so they are not re-proposed: MONDO:0022800 type 2 collagenopathy (contains spondylometaphyseal dysplasia 'corner fracture' type, listed in the SMD group); MONDO:0019695 acromelic dysplasia (contains the trichorhinophalangeal syndromes and Langer-Giedion, which the 2023 revision moved to group 19, plus terminal osseous dysplasia in group 6 and short-rib thoracic dysplasia 9 in the ciliopathy group — note the earlier rationale citing pseudohypoparathyroidism type 1A no longer applies, since the 2023 revision renamed that entity Albright hereditary osteodystrophy and placed it inside group 17; the class still straddles on the other three); MONDO:0017198 osteopetrosis (contains melorheostosis and osteopathia striata with cranial sclerosis, which belong to the osteosclerotic group); MONDO:0015338 syndromic craniosynostosis (contains cranioectodermal dysplasia, a skeletal ciliopathy). Several groups have no usable class at all — either the nearest MONDO term is obsolete (chondrodysplasia punctata) or it is *narrower* than the group, which mixes in entities falling outside it. In every mapped case MONDO remains broader than the group, since the nosology lists only entities meeting its inclusion criteria; that is why the relation is ``close_mappings`` and not ``exact_mappings``.

Classification of inborn errors of immunity based on the International Union of Immunological Societies (IUIS) Expert Committee classification. This is the gold-standard expert nosology for primary immunodeficiencies, updated biennially. Reference: Tangye et al. J Clin Immunol (2022) - IUIS 2022 Update

Biochemical classification of lysosomal storage diseases based on the type of accumulated substrate. Applies only to diseases caused by lysosomal enzyme deficiencies.

Classification of diseases by underlying molecular mechanism, affected pathway, or cellular structure. These "-opathy" categories group diseases that share defects in the same biological system regardless of clinical presentation.

Secondary, non-primary classification tagging a disease entry or curation project with the NIH "Highlighted Topics" funding-priority area(s) it is relevant to. This captures grant-strategy relevance, NOT disease nosology, so an entry may carry several tags or none. The topics are transient (each expires ~2 years after posting); this enum is GENERATED from a dated snapshot (2026-07-12) by scripts/gen_nih_topics_enum.py -- do not hand-edit.

Broad phenotype categories corresponding to the top-level organ-system groupings in the Human Phenotype Ontology (HPO). These are the direct children of HP:0000118 (Phenotypic abnormality) and are used to classify phenotypes by affected system.