The nature of anchor-biased estimates and its application to the wisdom of crowds.

We propose a method to achieve better wisdom of crowds by utilizing anchoring effects. In this method, people are first asked to make a comparative judgment such as "Is the number of new COVID-19 infections one month later more or less than 10 (or 200,000)?" As in this example, two sufficiently different anchors (e.g., "10" or "200,000") are set in the comparative judgment. After this comparative judgment, people are asked to make their own estimates. These estimates are then aggregated. We hypothesized that the aggregated estimates using this method would be more accurate than those without anchor presentation. To examine the effectiveness of the proposed method, we conducted three studies: a computer simulation and two behavioral experiments (numerical estimation of perceptual stimuli and estimation of new COVID-19 infections by physicians). Through computer simulations, we could identify situations in which the proposed method is effective. Although the proposed method is not always effective (e.g., when a group can make fairly accurate estimations), on average, the proposed method is more likely to achieve better wisdom of crowds. In particular, when a group cannot make accurate estimations (i.e., shows biases such as overestimation or underestimation), the proposed method can achieve better wisdom of crowds. The results of the behavioral experiments were consistent with the computer simulation findings. The proposed method achieved better wisdom of crowds. We discuss new insights into anchoring effects and methods for inducing diverse opinions from group members.

Data collaboration analysis in predicting diabetes from a small amount of health checkup data.

Recent studies showed that machine learning models such as gradient-boosting decision tree (GBDT) can predict diabetes with high accuracy from big data. In this study, we asked whether highly accurate prediction of diabetes is possible even from small data by expanding the amount of data through data collaboration (DC) analysis, a modern framework for integrating and analyzing data accumulated at multiple institutions while ensuring confidentiality. To this end, we focused on data from two institutions: health checkup data of 1502 citizens accumulated in Tsukuba City and health history data of 1399 patients collected at the University of Tsukuba Hospital. When using only the health checkup data, the ROC-AUC and Recall for logistic regression (LR) were 0.858 ± 0.014 and 0.970 ± 0.019, respectively, while those for GBDT were 0.856 ± 0.014 and 0.983 ± 0.016, respectively. When using also the health history data through DC analysis, these values for LR improved to 0.875 ± 0.013 and 0.993 ± 0.009, respectively, while those for GBDT deteriorated because of the low compatibility with a method used for confidential data sharing (although DC analysis brought improvements). Even in a situation where health checkup data of only 324 citizens are available, the ROC-AUC and Recall for LR were 0.767 ± 0.025 and 0.867 ± 0.04, respectively, thanks to DC analysis, indicating an 11% and 12% improvement. Thus, we concluded that the answer to the above question was "Yes" for LR but "No" for GBDT for the data set tested in this study.

DC-COX: Data collaboration Cox proportional hazards model for privacy-preserving survival analysis on multiple parties.

The demand for the privacy-preserving survival analysis of medical data integrated from multiple institutions or countries has been increased. However, sharing the original medical data is difficult because of privacy concerns, and even if it could be achieved, we have to pay huge costs for cross-institutional or cross-border communications. To tackle these difficulties of privacy-preserving survival analysis on multiple parties, this study proposes a novel data collaboration Cox proportional hazards (DC-COX) model based on a data collaboration framework for horizontally and vertically partitioned data. By integrating dimensionality-reduced intermediate representations instead of the original data, DC-COX obtains a privacy-preserving survival analysis without iterative cross-institutional communications or huge computational costs. DC-COX enables each local party to obtain an approximation of the maximum likelihood model parameter, the corresponding statistic, such as the p-value, and survival curves for subgroups. Based on a bootstrap technique, we introduce a dimensionality reduction method to improve the efficiency of DC-COX. Numerical experiments demonstrate that DC-COX can compute a model parameter and the corresponding statistics with higher performance than the local party analysis. Particularly, DC-COX demonstrates outstanding performance in essential feature selection based on the p-value compared with the existing methods including the federated learning-based method.

How the Content of Progress Notes Affects Readers' Perceptions of Their Usefulness and the Burden Placed on Writers: Improving Interprofessional Communication.

Electronic health records should efficiently store the information required for clinical decision-making and contain progress notes that reference this information. However, beyond the inclusion of subjective data, objective data, assessment, and plan framework, the content required to make progress notes useful for readers with diverse specialties has not been clarified. Moreover, the documentation burden that including additional content places on medical doctors (MDs) has not been determined. We conducted a questionnaire with 74 MDs, nurses, and other clinical professionals to determine whether they found progress notes with varying specific contents useful. In addition, the degree of the burden of writing progress notes that contain specific content was measured when 25 MDs were instructed to add specific content. Our results reveal that progress notes are more useful for clinical reasoning for readers other than MDs when more specific information is included; this can be achieved without increasing the documentation burden.

Validity of Lognormal Distribution in Analyzing Laboratory Test Values.

We compared the distribution of laboratory test values with several parametric statistical distributions to show that a lognormal distribution can represent the distribution of laboratory test values. Then, we estimated the distributions of laboratory test values of four datasets including only three published values: two endpoints of reference interval (RI) and one median.

On the round number bias and wisdom of crowds in different response formats for numerical estimation.

When asked for numerical estimations, people can respond by stating their estimates (e.g., writing down a number) or indicating a number on a scale. Although these methods are logically the same, such differences may affect the responses to the numerical estimations. In this study, we examined how differences in response format affected responses to numerical estimations using two behavioral experiments. We found that participants showed a round number bias (i.e., people answered estimates with round numbers) when simply stating a number and the distribution of responses tended to be less diverse. In contrast, this tendency was not observed when the participants responded using a scale. Participants provided more diverse estimates when they answered using a scale. Furthermore, we analyzed how this difference in response distribution was related to the wisdom of crowds (the aggregated judgment is as accurate as, or sometimes better than, the best individual judgment in the group) using computer simulations. The results indicated that round number bias affected the achievement of the wisdom of crowds. Particularly, when the group size was small, biased responses resulted in less effective achievement. Our findings suggest that using an appropriate scale is a low-cost method for eliminating round number bias and efficiently achieving the wisdom of crowds.

Automatic Classification of Electronic Nursing Narrative Records Based on Japanese Standard Terminology for Nursing.

In Japan, nursing records are not easily put to secondary use because nursing documentation is not standardized. In recent years, electronic health records have necessitated the creation of Japanese nursing terminology. The purpose of this study was to develop and evaluate an automatic classification system for narrative nursing records using natural language processing technology and machine learning. We collected a week's worth of narrative nursing records from an academic hospital. The authors independently annotated the text data, dividing it into morphemes, the smallest meaningful unit in a language. During preprocessing when creating feature quantities, we used a Japanese tokenizer, MeCab, an open-source morphological parser, and the bag-of-words model. A support vector machine was adopted as a classifier for machine learning. The accuracy was 0.96 and 0.86 on the training set and test set, respectively, and the F value was 0.82. Our findings provide useful information regarding the development of an automatic classification system for Japanese nursing records using nursing terminology and natural language processing techniques.

EHR2CCAS: A framework for mapping EHR to disease knowledge presenting causal chain of disorders - chronic kidney disease example.

OBJECTIVE: The goal of this work was to capture diseases in patients by comprehending the fine-grained medical conditions and disease progression manifested by transitions in medical conditions. We realize this by introducing our earlier work on a state-of-the-art knowledge presentation, which defines a disease as a causal chain of abnormal states (CCAS). Here, we propose a framework, EHR2CCAS, for constructing a system to map electronic health record (EHR) data to CCAS. MATERIALS AND METHODS: EHR2CCAS is a framework consisting of modules that access heterogeneous EHR to estimate the presence of abnormal states in a CCAS for a patient in a given time window. EHR2CCAS applies expert-driven (rule-based) and data-driven (machine learning) methods to identify abnormal states from structured and unstructured EHR data. It features data-driven approaches for unlocking clinical texts and imputations based on the EHR temporal properties and the causal CCAS structure. This study presents the CCAS of chronic kidney disease as an example. A mapping system between the EHR from the University of Tokyo Hospital and CCAS of chronic kidney disease was constructed and evaluated against expert annotation. RESULTS: The system achieved high prediction performance in identifying abnormal states that had strong agreement among annotators. Our handling of narrative varieties in texts and our imputation of the presence of an abnormal state markedly improved the prediction performance. EHR2CCAS presents patient data describing the temporal presence of abnormal states in CCAS, which is useful in individual disease progression management. Further analysis of the differentiation of transition among abnormal states outputted by EHR2CCAS can contribute to detecting disease subtypes. CONCLUSION: This work represents the first step toward combining disease knowledge and EHR to extract abnormality related to a disease defined as fine-grained abnormal states and transitions among them. This can aid in disease progression management and deep phenotyping.

Bias of Inaccurate Disease Mentions in Electronic Health Record-based Phenotyping.

OBJECTIVES: Electronic health record (EHR)-based phenotyping is an automated technique for identifying patients diagnosed with a particular disease using EHR data. However, EHR-based phenotyping has difficulties in achieving satisfactorily high performance because clinical notes include disease mentions that ultimately signify something other than the patient's diagnosis (such as differential diagnosis or screening). Our objective is to quantify the influence of such disease mentions on EHR-based phenotyping performance. METHODS: Physicians manually reviewed whether the disease mentions indicated the patients' diseases in 487,300 clinical notes of 4,430 patients. Particular focus was placed on disease mentions that did not signify the patient's diagnosis even though they did not have any syntactic modifier or indicator in the same sentences. Patients were then classified according to whether their clinical notes included such disease mentions. RESULTS: Among the patients whose clinical notes included disease mentions without any modifier or indicator, the proportion of patients whose disease mentions signified the patients' diagnosis was 78.1% (on average). This value can be interpreted as the bias of disease mentions that did not signify the patient's diagnosis on the precision of EHR-based phenotyping by extracting disease mentions from clinical notes. CONCLUSION: This study quantified the bias occurred owing to disease mentions that incorrectly signify a patient's diagnosis in the value of precision of EHR-based phenotyping from four dataset types. The results of this study will help researchers in diverse research environments with different available data types.

Development of Type 2 Diabetes Mellitus Phenotyping Framework Using Expert Knowledge and Machine Learning Approach.

BACKGROUND: Phenotyping is an automated technique that can be used to distinguish patients based on electronic health records. To improve the quality of medical care and advance type 2 diabetes mellitus (T2DM) research, the demand for T2DM phenotyping has been increasing. Some existing phenotyping algorithms are not sufficiently accurate for screening or identifying clinical research subjects. OBJECTIVE: We propose a practical phenotyping framework using both expert knowledge and a machine learning approach to develop 2 phenotyping algorithms: one is for screening; the other is for identifying research subjects. METHODS: We employ expert knowledge as rules to exclude obvious control patients and machine learning to increase accuracy for complicated patients. We developed phenotyping algorithms on the basis of our framework and performed binary classification to determine whether a patient has T2DM. To facilitate development of practical phenotyping algorithms, this study introduces new evaluation metrics: area under the precision-sensitivity curve (AUPS) with a high sensitivity and AUPS with a high positive predictive value. RESULTS: The proposed phenotyping algorithms based on our framework show higher performance than baseline algorithms. Our proposed framework can be used to develop 2 types of phenotyping algorithms depending on the tuning approach: one for screening, the other for identifying research subjects. CONCLUSIONS: We develop a novel phenotyping framework that can be easily implemented on the basis of proper evaluation metrics, which are in accordance with users' objectives. The phenotyping algorithms based on our framework are useful for extraction of T2DM patients in retrospective studies.

The Impact of "Possible Patients" on Phenotyping Algorithms: Electronic Phenotype Algorithms Can Only Be Reproduced by Sharing Detailed Annotation Criteria.

Phenotyping is an automated technique for identifying patients diagnosed with a particular disease based on electronic health records (EHRs). To evaluate phenotyping algorithms, which should be reproducible, the annotation of EHRs as a gold standard is critical. However, we have found that the different types of EHRs cannot be definitively annotated into CASEs or CONTROLs. The influence of such "possible patients" on phenotyping algorithms is unknown. To assess these issues, for four chronic diseases, we annotated EHRs by using information not directly referring to the diseases and developed two types of phenotyping algorithms for each disease. We confirmed that each disease included different types of possible patients. The performance of phenotyping algorithms differed depending on whether possible patients were considered as CASEs, and this was independent of the type of algorithms. Our results indicate that researchers must share annotation criteria for classifying the possible patients to reproduce phenotyping algorithms.