Handling missing data in large healthcare dataset: A case study of unknown trauma outcomes

Abstract

Handling of missed data is one of the main tasks in data preprocessing especially in large public service datasets. We have analysed data from the Trauma Audit and Research Network (TARN) database, the largest trauma database in Europe. For the analysis we used 165,559 trauma cases. Among them, there are 19,289 cases (11.35%) with unknown outcome. We have demonstrated that these outcomes are not missed 'completely at random' and, hence, it is impossible just to exclude these cases from analysis despite the large amount of available data. We have developed a system of non-stationary Markov models for the handling of missed outcomes and validated these models on the data of 15,437 patients which arrived into TARN hospitals later than 24h but within 30days from injury. We used these Markov models for the analysis of mortality. In particular, we corrected the observed fraction of death. Two naïve approaches give 7.20% (available case study) or 6.36% (if we assume that all unknown outcomes are 'alive'). The corrected value is 6.78%. Following the seminal paper of Trunkey (1983 [15]) the multimodality of mortality curves has become a much discussed idea. For the whole analysed TARN dataset the coefficient of mortality monotonically decreases in time but the stratified analysis of the mortality gives a different result: for lower severities the coefficient of mortality is a non-monotonic function of the time after injury and may have maxima at the second and third weeks. The approach developed here can be applied to various healthcare datasets which experience the problem of lost patients and missed outcomes.

Leveraging healthcare utilization to explore outcomes from musculoskeletal disorders: methodology for defining relevant variables from a health services data repository

“…The purpose of this paper is to use this surgical cohort as an example to describe the methodology employed to create meaningful data from large healthcare databases, and discuss relevant data considerations. In particular, we will discuss the methods utilized to address many of the ongoing challenges associated with use of big data, including: 1) sourcing data; 2) organizing data for clinical relevance [ 9 ]; 3) coding in a meaningful and descriptive way [ 10 ]; 4) handling missing values [ 11 – 14 ]; 5) reporting outcomes; 6) assuring the clinical veracity of the data [ 10 , 15 ]; and 7) reducing risks of analytic errors [ 16 ].…”

iHealthcare: Predictive Model Analysis Concerning Big Data Applications for Interactive Healthcare Systems †

“…An additional concern, while dealing with big data from the database, is to handle the missing attributes. According to academics, the Markov model could be a solution to handle missing data [3].…”

“…Some advanced technologies, such as tensors, cloud computing, and some intellectual frameworks, are used to analyze big data [3][4][5][6][7][8]. Big data processing, however, leads to problems, as much semi-structured and unstructured information exists.…”

Advanced Data Imputation Techniques for Predicting Type 2 Diabetes using Machine Learning

“…Data pre-processing is an important process to classify dataset accurately. In this context, Mirkes et al [21] proposed a modern approach to handle missing values and developed a system of Markov models for the handling of missing data and lost patients of Trauma Audit and Research Network (TARN). They have also imputed missing values with adjustment of weights on a dataset of patients of TARN.…”