SynTabFall

SynTabFall SynTabFall is a generated synthetic tabular dataset for fall prediction based on an anonymized German healthcare tabular dataset suitable for AI model development. The synthetic dataset counts 745,380 samples and 44 attributes, as shown in the schema below. The dataset consists mainly of attributes such as demographics, diseases, mobility and cognition related risk factors which are already used in practice. Column Dtype Values (if categorical) fallen bool - id object - age int64 - procedure object 1-20, 1-22, 1-24, 1-26 ... diagnosis object A011, A014, A020, A021 ... decubitus-admission bool - decubitus-at_the_moment bool - bed_mobility-impairment bool - bed_mobility-jones object 1, 2, 3, 4A, 4B bed_mobility-skin_condition bool - transfer-impairment bool - transfer object 1, 2, 3, 4A, 4B medical_items bool - medical_items-arm_or_leg_splint bool - medical_items-A_V_system bool - medical_items-plaster_or_neck_brace bool - medical_items-compression_stocking bool - medical_items-ventilation_mask bool - medical_items-O2_glasses bool - medical_items-orthesis bool - medical_items-prosthesis bool - medical_items-skin_condition_at_item_application bool - fall-last_12_month bool - fall-how_many_last_12_month float64 - fall-while_stay bool - fall-while_transfer bool - walk-impairment bool - walk-jones object 1, 2, 3, 4A, 4B walk-balance_and_gait_impaired bool - walking_aid bool - excretions-impairment bool - excretions-incontinence bool - excretions-nykturie bool - cognition-impairment bool - cognition-disoriented_time bool - cognition-disoriented_location bool - cognition-disoriented_own_person bool - cognition-agitated bool - cognition-confused bool - psychotropic_or_sedatives_drugs bool - decubitus-risk object +, ++, 0 fall-risk object +, ++, 0 diseases bool - sex object female, male Data Use Agreement (DUA) for SynTabFall This Data Use Agreement governs access to and use of the SynTabFall dataset. By accessing, downloading, or using the Dataset, you agree to comply with the terms below. 1. Purpose of Use (Permitted Use) You may use the Dataset solely for non-commercial research and educational purposes, including: scientific research and academic study, benchmarking and evaluation of machine learning models, reproducibility and verification of results reported in the associated publication(s). 2. Prohibited Use You agree not to: attempt to re-identify any real individual, attempt to link the dataset with external datasets for re-identification purposes, use the dataset for clinical decision-making, medical diagnosis, or patient care, use the dataset for commercial purposes without explicit written permission, redistribute, re-host, or publicly share the dataser in full or in part, unless explicitly authorized by the data provider. 3. Privacy and Re-identification Restrictions Although the Dataset is synthetic, it is derived from sensitive clinical data.Users must not attempt to infer whether a specific individual was part of the original training data. Users agree not to perform, publish, or release results from: attribute inference attacks, membership inference attacks, linkage attacks, singling-out attacks. when the goal is identification or inference about real individuals. Security/privacy evaluations are permitted only when performed ethically and reported in aggregate form, without attempting to identify any person. 4. Data Security Requirements Users agree to take reasonable technical and organizational measures to prevent unauthorized access to the dataset, including: storing the dataset on secure local or institutional infrastructure, limiting access to authorized researchers, avoiding uploading the dataset to public repositories or public cloud storage without access controls 5. Citation and Attribution If you use the Dataset in a publication, presentation, or derived work, you must cite the associated dataset paper. 6. Derived Work and Redistribution Users may publish: trained models, aggregated statistics, evaluation results, plots and tables provided that the dataset itself (or a substantial portion of it) is not redistributed. Redistribution of the dataset, whether modified or unmodified, requires explicit permission from the Data Providers. 7. Disclaimer The dataset is provided “as is”, without warranty of any kind.The data providers make no guarantees regarding correctness, completeness, or suitability for any purpose. The dataset must not be used for medical diagnosis, treatment decisions, or any clinical application. 8. Termination Violation of this Agreement may result in termination of access and the requirement to delete all copies of the dataset. 9. Acceptance By accessing, downloading, or using the dataset, you confirm that you have read, understood, and agreed to comply with this Agreement.