The company we keep. Using hemodialysis social network data to classify patients' kidney transplant attitudes with machine learning algorithms

Abstract

Background: Hemodialysis clinic patient social networks may reinforce positive and negative attitudes towards kidney transplantation. We examined whether a patient's position within the hemodialysis clinic social network could improve machine learning classification of the patient's positive or negative attitude towards kidney transplantation when compared to sociodemographic and clinical variables.

Methods: We conducted a cross-sectional social network survey of hemodialysis patients in two geographically and demographically different hemodialysis clinics. We evaluated whether machine learning logistic regression models using sociodemographic or network data best predicted the participant's transplant attitude. Models were evaluated for accuracy, precision, recall, and F1-score.

Results: The 110 surveyed participants' mean age was 60 ± 13 years old. Half (55%) identified as male, and 74% identified as Black. At facility 1, 69% of participants had a positive attitude towards transplantation whereas at facility 2, 45% of participants had a positive attitude. The machine learning logistic regression model using network data alone obtained a higher accuracy and F1 score than the sociodemographic and clinical data model (accuracy 65% ± 5% vs. 61% ± 7%, F1 score 76% ± 2% vs. 70% ± 7%). A model with a combination of both sociodemographic and network data had a higher accuracy of 74% ± 3%, and an F1-score of 81% ± 2%.

Conclusion: Social network data improved the machine learning algorithm's ability to classify attitudes towards kidney transplantation, further emphasizing the importance of hemodialysis clinic social networks on attitudes towards transplant.

Keywords: Hemodialysis; Kidney transplantation; Machine Learning; Psychosocial; Social Determinants of Health; Social Network; Survey Research.

Fig. 1

Social-ecological model for kidney transplant disparities. Figure 1. Social-ecological model for kidney transplant disparities is a modified version of two models. [7, 8] The first layer is the individual layer which refers to the patients knowledge, attitudes, behaviors, and biology. This layer is shaped by the other layers of the model. Such as the interpersonal layer (i.e. the individual’s social network), the institutional layer (e.g. the healthcare system), the community layer (e.g. the culture of organ donation and transplant with in the community, the public policy layer (e.g. mandated transplant education). All these layers influence each other and ultimately shape the individual’s beliefs

Fig. 2

Glossary of Terms. Figure 2. presents a glossary of terms as well as a diagram of a kite network used to demonstrate different centrality measures. Each dot represents a person in the network and they are labelled a-i. A black line represents a relationship between two members of the network

Fig. 3

Network graphs of the hemodialysis clinics. Figure 3. Network graphs of the hemodialysis clinics. The green circles (nodes) represent participants with a positive attitude towards transplant and the red nodes represent patients with a negative attitude towards transplant. A blue line (edge) between the participants represents a relationship. The Monday, Wednesday, Friday (MWF) and Tuesday, Thursday, Saturday (TTS) shifts are circled. Note at facility 1 there were no relationships that spanned the MWF and TTS whereas there were two relationships that spanned the MWF and TTS at facility 2

Fig. 4

Random Classifier Reciever Operator Curve of Combined Logistic Regression Model. Figure 4. shows the receiver operator of the machine learning combined network, sociodemographic, and clinical data with false positive rate on the x-axis and true positive rate on the the y-axis. LR (logistic regression). [38]