Explainable AI for Early Detection of Health Changes Via Streaming Clustering

The ability to explain the predictions of machine learning models has become increasingly important, especially in healthcare applications. Streaming clustering is an effective tool to recognize normal baseline patterns and to detect early signs of changes in data streams. However, many streaming cl...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE International Fuzzy Systems conference proceedings s. 1 - 6
Hlavní autoři: Wu, Wenlong, Keller, James M., Skubic, Marjorie, Popescu, Mihail
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 18.07.2022
Témata:
Annotations
Behavioral sciences
Clustering algorithms
explainable ai
Linguistics
Machine learning
Prediction algorithms
Predictive models
streaming clustering
ISSN:1558-4739
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:The ability to explain the predictions of machine learning models has become increasingly important, especially in healthcare applications. Streaming clustering is an effective tool to recognize normal baseline patterns and to detect early signs of changes in data streams. However, many streaming clustering algorithms are not designed to explain to the users how predictions are made. In this paper, we extend a streaming clustering algorithm, the sequential possibilistic Gaussian mixture model (SPGMM) for early detection of health change to provide algorithm explainability for the results. Four approaches are discussed to explain either the cluster differences or the reason for the algorithm warnings: (i) linguistic summarization for warnings; (ii) annotation distribution of clusters; (iii) SHapley Additive exPlanations (SHAP); (iv) functional health score. The four approaches are validated on one older adult monitored with a collection of motion, bed, and depth sensors over three years. The results obtained on the older adult show that the four approaches aid understanding of how the clusters and warnings are generated, providing strong support for clinicians to take corresponding actions.
ISSN:1558-4739
DOI:10.1109/FUZZ-IEEE55066.2022.9882813