The increasing availability of electronic medical records makes it possible to reconstruct patient treatment patterns adopted in a given clinical setting. Developing methods to detect deviations from these patterns may help to determine whether the management of a patient is unusual in some way. Using pattern mining techniques, our method extracts frequent patterns from a given dataset of treatment undergone by patients. Comparison is then made between frequent patterns of treatment and the domain medical knowledge, thus allowing the detection of two types of anomalies. The first type includes frequent patterns which deviate from accepted guidelines. These patterns can be evaluated and eventually fed back into improving the guidelines. The second type of anomalies comprises the anomalous cases that deviate from the frequent patterns. They may simply indicate variations in the examinations prescribed due to specific patient conditions, otherwise they may reveal limitation in accessing public health services or identify errors in the data entry process. In all these cases, the detection of the anomalies is useful for a successive analysis by domain experts. We applied our method to three case studies to show how it can be successfully exploited in real medical domain.
Anomaly detection in medical treatment to discover unusual patient management / Antonelli, Dario; Bruno, Giulia; Chiusano, SILVIA ANNA. - In: IIE TRANSACTIONS ON HEALTHCARE SYSTEMS ENGINEERING. - ISSN 1948-8300. - 3:(2013), pp. 69-77. [10.1080/19488300.2013.787564]
Anomaly detection in medical treatment to discover unusual patient management
ANTONELLI, DARIO;BRUNO, GIULIA;CHIUSANO, SILVIA ANNA
2013
Abstract
The increasing availability of electronic medical records makes it possible to reconstruct patient treatment patterns adopted in a given clinical setting. Developing methods to detect deviations from these patterns may help to determine whether the management of a patient is unusual in some way. Using pattern mining techniques, our method extracts frequent patterns from a given dataset of treatment undergone by patients. Comparison is then made between frequent patterns of treatment and the domain medical knowledge, thus allowing the detection of two types of anomalies. The first type includes frequent patterns which deviate from accepted guidelines. These patterns can be evaluated and eventually fed back into improving the guidelines. The second type of anomalies comprises the anomalous cases that deviate from the frequent patterns. They may simply indicate variations in the examinations prescribed due to specific patient conditions, otherwise they may reveal limitation in accessing public health services or identify errors in the data entry process. In all these cases, the detection of the anomalies is useful for a successive analysis by domain experts. We applied our method to three case studies to show how it can be successfully exploited in real medical domain.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/11583/2507673
Attenzione
Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo