intelmedica/nursing-sentences-1

--- license: cc-by-nc-4.0 language: - en task_categories: - text-generation - automatic-speech-recognition tags: - medical - clinical - nursing - synthetic - asr-training - sentence-generation - clinical-documentation pretty_name: "IntelMedica Nursing Sentences v1" size_categories: - 10K<n<100K dataset_info: features: - name: text dtype: string - name: category dtype: string - name: source_api dtype: string - name: term dtype: string - name: audience dtype: string splits: - name: train num_examples: 28173 - name: validation num_examples: 6037 - name: test num_examples: 6037 --- # IntelMedica Nursing Sentences v1 Synthetic nursing-specific clinical documentation sentences for training medical Automatic Speech Recognition (ASR) models. Part of the [IntelMedica](https://intelmedica.ai) open-source medical AI initiative. ## Overview | Stat | Value | |------|-------| | **Total rows** | 40,247 | | **Train** | 28,173 | | **Validation** | 6,037 | | **Test** | 6,037 | | **Split ratio** | 70 / 15 / 15 (stratified by category) | | **Language** | English | | **Audience** | Nursing | ## Category Distribution | Category | Train | Val | Test | Total | |----------|------:|----:|-----:|------:| | sbar | 6,920 | — | — | ~9,886 | | med_admin | 6,662 | — | — | ~9,517 | | nursing_assessment | 4,385 | — | — | ~6,264 | | wound_care | 2,964 | — | — | ~4,234 | | pain | 2,370 | — | — | ~3,386 | | vitals | 2,331 | — | — | ~3,330 | | vitals_assessment_combo | 998 | — | — | ~1,426 | | drug_side_effect_combo | 579 | — | — | ~827 | | intake_output | 362 | — | — | ~517 | | safety | 317 | — | — | ~453 | *16 categories total. Counts shown for train split; val/test follow same distribution.* ## Schema | Column | Type | Description | |--------|------|-------------| | `text` | string | The generated clinical sentence | | `category` | string | Clinical documentation category (e.g., sbar, hpi, soap_assessment) | | `source_api` | string | Origin API of the medical term used in generation | | `term` | string | The medical term the sentence was built around | | `audience` | string | Target audience: `nursing` | ## Data Sources Medical terms were collected from 11+ authoritative APIs and databases: | Source | Terms | Notes | |--------|------:|-------| | cross_source | 37,410 | Multi-API combined terms | | combined | 2,252 | Merged from multiple sources | | nursing_curated | 366 | Hand-curated nursing terms | | nursing_physician | 216 | Cross-audience nursing/physician terms | | abbreviations | 2 | Medical abbreviations | | snomed_ct | 1 | SNOMED CT terms | ## Generation Pipeline 1. **Term collection** from 11 medical terminology APIs (RxNorm, SNOMED CT, NCI Thesaurus, MeSH, LOINC, DailyMed, HCPCS, FDA, CMS, plus curated nursing terms and 104K medical abbreviations) 2. **Quality cleaning** with 12 rules (deduplication, length filtering, encoding fixes, garbage removal) -- removed ~10% low-quality entries 3. **Template-based sentence generation** using Qwen 3.5 2B with audience-specific templates (nursing clinical scenarios) 4. **Stratified splitting** into 70/15/15 train/validation/test by category Full pipeline code: [intelmedica/med-speech-data-prep](https://github.com/intelmedica/med-speech-data-prep) ## Audio Versions Audio versions (TTS-synthesized at 16kHz, multi-speaker) coming soon: - `intelmedica/medical-tts-nursing-16khz` - `intelmedica/medical-tts-physician-16khz` - `intelmedica/medical-tts-general-16khz` ## Usage ```python from datasets import load_dataset ds = load_dataset("intelmedica/nursing-sentences-1") print(ds) # DatasetDict({ # train: Dataset({features: ['text', 'category', 'source_api', 'term', 'audience'], num_rows: 28173}) # validation: Dataset({features: [...], num_rows: 6037}) # test: Dataset({features: [...], num_rows: 6037}) # }) print(ds["train"][0]) ``` ## Related Datasets - [jfmdai/medical-speech-data-collections](https://huggingface.co/datasets/jfmdai/medical-speech-data-collections) -- Field directory of all medical speech datasets - [jfmdai/nursing-sentences](https://huggingface.co/datasets/jfmdai/nursing-sentences) -- Original source (nursing) - [jfmdai/physician-sentences](https://huggingface.co/datasets/jfmdai/physician-sentences) -- Original source (physician) - [jfmdai/general-medical-sentences](https://huggingface.co/datasets/jfmdai/general-medical-sentences) -- Original source (general) ## Why `-1`? This is **version 1**. Future versions will incorporate: - Additional APIs (PubMed, RadLex, ClinicalTrials.gov) - Accent diversity via voice cloning - LLM-generated contextual clinical scenarios - Real-world correction-based improvements from deployed ASR systems ## License [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/) ## Citation ```bibtex @dataset{nursing_sentences_1, author = {Farooq, Junaid}, title = {IntelMedica Nursing Sentences v1}, year = {2026}, publisher = {Hugging Face}, url = {https://huggingface.co/datasets/intelmedica/nursing-sentences-1}, note = {Synthetic medical sentences for ASR training} } ``` ## Author **Junaid Farooq, MD** / [IntelMedica LLC](https://intelmedica.ai) / Physician-Led Open-Source Medical AI ## Disclaimer This dataset is for **research purposes only**. It is not a medical device, not Software as a Medical Device (SaMD), and not intended for clinical decision support. All data is **synthetic** -- no Protected Health Information (PHI) is present. Generated from publicly available medical terminology databases.--- 许可证：CC BY-NC 4.0 语言： - 英语 任务类别： - 文本生成 - 自动语音识别（Automatic Speech Recognition, ASR） 标签： - 医疗 - 临床 - 护理 - 合成 - ASR训练 - 句子生成 - 临床文档 展示名称："IntelMedica 护理语句 v1" 规模类别： - 10000 < 样本数 < 100000 数据集信息： 特征： - 字段名：text，数据类型：字符串 - 字段名：category，数据类型：字符串 - 字段名：source_api，数据类型：字符串 - 字段名：term，数据类型：字符串 - 字段名：audience，数据类型：字符串 数据集划分： - 划分名称：train，样本数量：28173 - 划分名称：validation，样本数量：6037 - 划分名称：test，样本数量：6037 --- # IntelMedica 护理语句 v1 用于训练医疗自动语音识别（Automatic Speech Recognition, ASR）模型的专用合成护理临床文档语句。属于[IntelMedica](https://intelmedica.ai)开源医疗人工智能（AI）计划的一部分。 ## 概览 | 统计项 | 数值 | |------|-------| | **总样本数** | 40,247 | | **训练集** | 28,173 | | **验证集** | 6,037 | | **测试集** | 6,037 | | **划分比例** | 70 / 15 / 15（按类别分层） | | **语言** | 英语 | | **目标受众** | 护理人员 | ## 类别分布 | 类别 | 训练集 | 验证集 | 测试集 | 总计 | |----------|------:|----:|-----:|------:| | sbar | 6,920 | — | — | ~9,886 | | med_admin | 6,662 | — | — | ~9,517 | | nursing_assessment | 4,385 | — | — | ~6,264 | | wound_care | 2,964 | — | — | ~4,234 | | pain | 2,370 | — | — | ~3,386 | | vitals | 2,331 | — | — | ~3,330 | | vitals_assessment_combo | 998 | — | — | ~1,426 | | drug_side_effect_combo | 579 | — | — | ~827 | | intake_output | 362 | — | — | ~517 | | safety | 317 | — | — | ~453 | *总计16个类别。表格中仅展示训练集样本量；验证集与测试集遵循相同的类别分布。* ## 数据结构 | 字段 | 数据类型 | 说明 | |--------|------|-------------| | `text` | 字符串 | 生成的临床语句 | | `category` | 字符串 | 临床文档类别（例如sbar、hpi、soap_assessment） | | `source_api` | 字符串 | 生成所用医疗术语的来源API | | `term` | 字符串 | 构建语句所围绕的医疗术语 | | `audience` | 字符串 | 目标受众：`nursing`（护理人员） | ## 数据来源 医疗术语收集自11个以上权威API与数据库： | 来源 | 术语数量 | 备注 | |--------|------:|-------| | cross_source | 37,410 | 多API组合术语 | | combined | 2,252 | 多源合并术语 | | nursing_curated | 366 | 人工整理的护理术语 | | nursing_physician | 216 | 跨受众的护理/医师术语 | | abbreviations | 2 | 医疗缩写 | | snomed_ct | 1 | SNOMED CT术语 | ## 生成流程 1. **术语收集**：从11个医疗术语API（包括RxNorm、SNOMED CT、美国国家癌症研究所词表（National Cancer Institute Thesaurus, NCI Thesaurus）、医学主题词表（Medical Subject Headings, MeSH）、观测信息与数据标准逻辑命名（Logical Observation Identifiers Names and Codes, LOINC）、DailyMed、医疗保健通用程序编码系统（Healthcare Common Procedure Coding System, HCPCS）、美国食品药品监督管理局（Food and Drug Administration, FDA）、美国医疗保险和医疗补助服务中心（Centers for Medicare & Medicaid Services, CMS），外加人工整理的护理术语与10.4万个医疗缩写）获取术语 2. **质量清洗**：通过12项规则（去重、长度过滤、编码修复、垃圾内容清除）处理，移除了约10%的低质量条目 3. **基于模板的语句生成**：使用Qwen 3.5 2B模型，结合面向受众的模板（护理临床场景）生成语句 4. **分层划分**：按类别以70/15/15的比例划分为训练集、验证集与测试集 完整流程代码：[intelmedica/med-speech-data-prep](https://github.com/intelmedica/med-speech-data-prep) ## 语音版本 16kHz、多说话人合成的语音版本（文本转语音（Text-to-Speech, TTS）生成）即将推出： - `intelmedica/medical-tts-nursing-16khz` - `intelmedica/medical-tts-physician-16khz` - `intelmedica/medical-tts-general-16khz` ## 使用方法 python from datasets import load_dataset ds = load_dataset("intelmedica/nursing-sentences-1") print(ds) # DatasetDict({ # train: Dataset({features: ['text', 'category', 'source_api', 'term', 'audience'], num_rows: 28173}) # validation: Dataset({features: [...], num_rows: 6037}) # test: Dataset({features: [...], num_rows: 6037}) # }) print(ds["train"][0]) ## 相关数据集 - [jfmdai/medical-speech-data-collections](https://huggingface.co/datasets/jfmdai/medical-speech-data-collections) —— 所有医疗语音数据集的领域目录 - [jfmdai/nursing-sentences](https://huggingface.co/datasets/jfmdai/nursing-sentences) —— 原始来源（护理方向） - [jfmdai/physician-sentences](https://huggingface.co/datasets/jfmdai/physician-sentences) —— 原始来源（医师方向） - [jfmdai/general-medical-sentences](https://huggingface.co/datasets/jfmdai/general-medical-sentences) —— 原始来源（通用医疗方向） ## 为何后缀为`-1`？ 此为**版本1**。未来版本将纳入： - 更多API（PubMed、RadLex、ClinicalTrials.gov） - 基于语音克隆的口音多样性支持 - 大语言模型（Large Language Model, LLM）生成的上下文临床场景 - 来自部署的ASR系统的真实世界修正优化 ## 许可证 [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/) ## 引用 bibtex @dataset{nursing_sentences_1, author = {Farooq, Junaid}, title = {IntelMedica Nursing Sentences v1}, year = {2026}, publisher = {Hugging Face}, url = {https://huggingface.co/datasets/intelmedica/nursing-sentences-1}, note = {用于ASR训练的合成医疗语句} } ## 作者 **Junaid Farooq, MD** / [IntelMedica LLC](https://intelmedica.ai) / 医师主导的开源医疗AI团队 ## 免责声明 本数据集仅用于**研究目的**。它不属于医疗设备，也不属于软件即医疗器械（Software as a Medical Device, SaMD），不得用于临床决策支持。所有数据均为**合成数据**，不包含任何受保护健康信息（Protected Health Information, PHI），均生成自公开可用的医疗术语数据库。