Lab indicators standardization method for the regional healthcare platform: a case study on heart failure

Abstract

Background Laboratory indicator test results in electronic health records have been applied to many clinical big data analysis. However, it is quite common that the same laboratory examination item (i.e., lab indicator) is presented using different names in Chinese due to the translation problem and the habit problem of various hospitals, which results in distortion of analysis results. Methods A framework with a recall model and a binary classification model is proposed, which could reduce the alignment scale and improve the accuracy of lab indicator normalization. To reduce alignment scale, tf-idf is used for candidate selection. To assure the accuracy of output, we utilize enhanced sequential inference model for binary classification. And active learning is applied with a selection strategy which is proposed for reducing annotation cost. Results Since our indicator standardization method mainly focuses on Chinese indicator inconsistency, we perform our experiment on Shanghai Hospital Development Center and select clinical data from 8 hospitals. The method achieves a F1-score 92.08$$\%$$ % in our final binary classification. As for active learning, the new strategy proposed performs better than random baseline and could outperform the result trained on full data with only 43$$\%$$ % training data. A case study on heart failure clinic analysis conducted on the sub-dataset collected from SHDC shows that our proposed method is practical in the application with good performance. Conclusion This work demonstrates that the structure we proposed can be effectively applied to lab indicator normalization. And active learning is also suitable for this task for cost reduction. Such a method is also valuable in data cleaning, data mining, text extracting and entity alignment.

Informatics and machine learning methods for health applications

“…Most of the studies included here were facilitated by and conducted with the valuable data resources either available in the open science domain or accessible to the authors. In five of these articles [ 1 , 4 , 7 , 9 , 10 ], electronic health record (EHR) data were analyzed for various purposes, including (1) predicting outcomes such as mortality [ 1 , 7 , 10 ], sepsis [ 7 ], and preterm birth [ 10 ], (2) annotation and extraction of age and temporally-related events [ 4 ], and (3) sepsis phenotyping [ 9 ]. In the remaining five articles [ 2 , 3 , 5 , 6 , 8 ], the authors analyzed (1) publication data for extracting biomedical concepts [ 2 ], (2) wearable sensor data for stress detection [ 3 ], (3) location data for resource allocation for cardiac emergency [ 5 ], (4) social media data for drug use detection [ 6 ], and (5) ECG data for biomedical signal denoising [ 8 ].…”

“…In these studies, the authors employed informatics and machine learning methods to address various health topics, including diabetes [ 1 ], autism spectrum disorder [ 2 ], stress [ 3 ], health research in general [ 4 ], cardiac arrest [ 5 ], drug use [ 6 ], sepsis [ 7 , 9 ], heart disorders [ 8 ], and preterm birth and perinatal mortality [ 10 ]. To address the biomedical problems in the above health applications, these studies employed a wide range of informatics and machine learning methods, including deep learning [ 1 , 3 , 6 , 7 ], NLP [ 1 , 2 , 4 ], matching algorithms [ 5 ], association mining [ 6 ], wavelet analysis [ 8 ], factor analysis [ 9 ], frequent graph mining [ 9 ], and traditional statistical machine learning [ 10 ].…”

“…In “ Predicting mortality in critically ill patients with diabetes using machine learning and clinical notes ”, Ye et al [ 1 ] investigated the mortality rate of patients with diabetes admitted to intensive care unit (ICU), using machine learning and clinical notes, while previous work was mostly based on regression models and did not take clinical notes into considerations. Specifically, the authors proposed to use Unified Medical Language System (UMLS) resources, coupled with machine learning and natural language processing (NLP) approaches, to predict the risk of mortality.…”

“…These ten articles went through the second round of peer-review and revision in the BMC Supplement submission system before their final acceptance to this ICIBM 2020 supplement issue. Of note, eight articles [ 1 – 8 ] are included here (Part I of the special issue), while the other two articles [ 9 , 10 ] will be included in the Part II of the special issue scheduled to appear in 2021. Below we briefly summarize these ten selected manuscripts.…”

A multi-perspective combined recall and rank framework for Chinese procedure terminology normalization

“…With the size of knowledge base increases, point-wise learning to rank methods faces serious efficiency problem. Rank follows recall approaches [14][15][16][17][18] applies a two-step framework which first generate candidate terminologies by heuristic rules or statistic methods then rank the candidates in a point-wise way. The recall step only decreases the scale of the candidates without providing other information for sorting the candidates.…”