The effect of disease co-occurrence measurement on multimorbidity networks
| dc.contributor.author | Monchka, Barret | |
| dc.contributor.examiningcommittee | Nickel, Nathan (Community Health Sciences) | en_US |
| dc.contributor.examiningcommittee | Leung, Carson (Computer Science) | en_US |
| dc.contributor.supervisor | Lix, Lisa (Community Health Sciences) | en_US |
| dc.date.accessioned | 2021-08-31T20:19:37Z | |
| dc.date.available | 2021-08-31T20:19:37Z | |
| dc.date.copyright | 2021-08-25 | |
| dc.date.issued | 2021 | en_US |
| dc.date.submitted | 2021-08-25T17:07:07Z | en_US |
| dc.degree.discipline | Community Health Sciences | en_US |
| dc.degree.level | Master of Science (M.Sc.) | en_US |
| dc.description.abstract | Background: Network analysis, a technique for describing relationships, can provide insights into patterns of co-occurring chronic diseases. The effect that co-occurrence measurement has on disease network structure and resulting inferences has not been well studied. Objectives: The research objectives were to (1) compare structural differences among chronic disease networks constructed from different co-occurrence measures, and (2) demonstrate how co-occurrences among three or more chronic diseases can be analyzed using network techniques. Methods: A retrospective cohort study was conducted using four years of Manitoba administrative health data (2015/16 – 2018/19, 1.5 million individuals). Association rule mining was used to identify disease triads. Separate disease networks were constructed using seven co-occurrence measures: lift, relative risk, phi, Jaccard, cosine, Kulczynski, and joint prevalence. Influential diseases were identified using degree centrality and community detection was used to identify disease clusters. Community structure similarity was measured using the adjusted Rand index (ARI). Network edges were described using disease prevalence categorized as low (<1%), moderate (1% to <7%), and high (≥7%). Results: Relative risk and lift highlighted co-occurrences between pairs of low prevalent diseases. Kulczynski emphasized relationships between high and low prevalent diseases. Joint prevalence focused on highly prevalent conditions. Phi, Jaccard, and cosine emphasized associations with moderately prevalent conditions. Co-occurrence measurement differences significantly affected how disease clusters were defined, including the number of clusters identified. When limiting the number of edges to produce visually interpretable graphs, networks had significant dissimilarity in the percentage of co-occurrence relationships in common, and in their selection of the highest degree nodes. Conclusion: Multimorbidity network analyses are sensitive to disease co-occurrence measurement. Co-occurrence measures should be selected considering research objectives and the prevalence relationships of greatest interest. Researchers should be cautious in their interpretation of findings from network analysis and should conduct sensitivity analyses using different co-occurrence measures. Many chronic diseases co-occur in groups of three or more and these higher-order associations can be visualized and analyzed using hypergraphs. | en_US |
| dc.description.note | October 2021 | en_US |
| dc.identifier.uri | http://hdl.handle.net/1993/35871 | |
| dc.language.iso | eng | en_US |
| dc.rights | open access | en_US |
| dc.subject | Multimorbidity | en_US |
| dc.subject | Chronic disease | en_US |
| dc.subject | Network analysis | en_US |
| dc.subject | Canada | en_US |
| dc.subject | Manitoba | en_US |
| dc.subject | Disease network | en_US |
| dc.subject | Hypergraph | en_US |
| dc.title | The effect of disease co-occurrence measurement on multimorbidity networks | en_US |
| dc.type | master thesis | en_US |
| local.subject.manitoba | yes | en_US |