A classification for complex imbalanced data in disease screening and early diagnosis

Abstract

Imbalanced classification has drawn considerable attention in the statistics and machine learning literature. Typically, traditional classification methods often perform poorly when a severely skewed class distribution is observed, not to mention under a high-dimensional longitudinal data structure. Given the ubiquity of big data in modern health research, it is expected that imbalanced classification in disease diagnosis may encounter an additional level of difficulty that is imposed by such a complex data structure. In this article, we propose a nonparametric classification approach for imbalanced data in longitudinal and high-dimensional settings. Technically, the functional principal component analysis is first applied for feature extraction under the longitudinal structure. The univariate exponential loss function coupled with group LASSO penalty is then adopted into the classification procedure in high-dimensional settings. Along with a good improvement in imbalanced classification, our approach provides a meaningful feature selection for interpretation while enjoying a remarkably lower computational complexity. The proposed method is illustrated on the real data application of Alzheimer's disease early detection and its empirical performance in finite sample size is extensively evaluated by simulations.

Differentiating Pressure Ulcer Risk Levels Using Decision Trees Based on Interpretable Indicators: A Practical Approach for Nursing Care

“…The model is built using a decision tree generator algorithm [16,21], whose output is easily interpretable and applicable in a multitude of contexts. Data preprocessing: As shown in Table 1, the dataset obtained is characterized by being imbalanced [14]. This fact indicates that the data present an unequal distribution among classes [15], implying that the percentage of observations at each PUs risk level is quite uneven.…”

“…This, in turn, will facilitate effective interventions, including the selection of support surfaces and other relevant applications. For this purpose, data from 16,215 patients in Granada (Spain) have been collected, focusing mainly on indicators that are quickly identifiable in practice by nursing professionals, such as mobility, activity and skin humidity [4] The data treated present a common problem in nursing classification datasets, such as the presence of class imbalance [14]. In order address this issue, the dataset is preprocessed and various oversampling configurations are studied to determine the best one allowing to improve data quality before modeling [15].…”

Differentiating Pressure Ulcer Risk Levels through Interpretable Classification Models Based on Readily Measurable Indicators

“…Data preprocessing. As shown in Table 1, the dataset obtained is characterized by being imbalanced [26]. This fact indicates that the data present an unequal distribution among classes [27], implying that the percentage of observations at each PU risk level is quite uneven.…”

“…For this purpose, data from 16,215 patients in Granada (Spain) have been collected, focusing mainly on indicators that are quickly identifiable in practice by nursing professionals, such as mobility, activity, and skin humidity [4]. The data present a common problem in nursing classification datasets, such as the presence of class imbalance [26]. In order to address this issue, the dataset is preprocessed and various oversampling configurations are studied to determine the best one to enable the improvement of the data quality before modeling [27].…”

Imbalanced Data Over-Sampling Method Based on ISODATA Clustering

“…For example, in the medical diagnosis of human cancer disease compared to healthy people belong to the minority, but obviously, people pay more attention to cancer disease. If the patient is diagnosed as healthy, the patient will not receive timely treatment, resulting in irreparable results [7], [8]. Therefore, the analysis of minority data is particularly important.…”