Metabolic syndrome is a cluster of conditions that increase the risk of cardiovascular disease, stroke, and type 2 diabetes. This thesis aims to predict the onset of metabolic syndrome in the population of blood donors. The data was provided by AVIS Milano and includes variables related to the donor’s lifestyle and the blood test results for each do nation. After pre-processing the data we fit it in a two-stage plug-in model. For the first stage, we considered three mixed-effect Bayesian models for longitudinal data to predict future values of the five responses defining the metabolic syndrome. These results have been plugged into a logistic model to estimate the donor’s overall probability of developing metabolic syndrome. Three risk zones were identified to classify the donors and we were able to correctly identify up to 90% of at-risk donors. As a result of this thesis, a screening tool was developed and provided to AVIS. The posterior analysis of the Bayesian models reveals a strong correlation between metabolic syndrome and lifestyle-related variables such as BMI, physical activity levels, and smoking habits. The final classification results are very promising and will allow AVIS to more effectively identify at-risk donors before the onset of metabolic syndrome, optimizing resources and improving donors’ health.
La sindrome metabolica è un insieme di condizioni che aumentano il rischio di malattie cardiovascolari, ictus e diabete di tipo 2. Questa tesi mira a prevedere l’insorgenza della sindrome metabolica nella popolazione dei donatori di sangue. I dati sono stati forniti da AVIS Milano e includono variabili relative allo stile di vita dei donatori e i risultati delle analisi del sangue di ogni donazione. Dopo aver pulito i dati sono stati inseriti in un modello plug-in a due fasi. Nella prima fase, abbiamo considerato tre modelli a effetti misti bayesiani per dati longitudinali al fine di predirre i valori futuri delle cinque vari abili necessarie per diagnosticare la sindrome metabolica. Questi risultati sono stati poi utilizzati in un modello logistico per stimare, per ogni donatore, la probabilità di svilup pare la sindrome metabolica. Sono state identificate tre zone di rischio in cui classificare i donatori, e siamo stati in grado di identificare correttamente fino al 90% dei donatori a rischio. Come risultato di questa tesi, è stato sviluppato uno strumento di screening fornito ad AVIS. L’analisi a posteriori dei modelli bayesiani ha rivelato una forte corre lazione tra la sindrome metabolica e le variabili legate allo stile di vita come il BMI, i livelli di attività fisica e le abitudini di fumo. I risultati finali della classificazione sono molto promettenti e permetteranno ad AVIS di identificare più efficacemente i donatori a rischio prima dell’insorgenza della sindrome metabolica, ottimizzando così le risorse e migliorando la salute dei donatori.
Bayesian models for early diagnosis and prediction of metabolic syndrome in healthy blood donors
Arrigoni, Francesca
2023/2024
Abstract
Metabolic syndrome is a cluster of conditions that increase the risk of cardiovascular disease, stroke, and type 2 diabetes. This thesis aims to predict the onset of metabolic syndrome in the population of blood donors. The data was provided by AVIS Milano and includes variables related to the donor’s lifestyle and the blood test results for each do nation. After pre-processing the data we fit it in a two-stage plug-in model. For the first stage, we considered three mixed-effect Bayesian models for longitudinal data to predict future values of the five responses defining the metabolic syndrome. These results have been plugged into a logistic model to estimate the donor’s overall probability of developing metabolic syndrome. Three risk zones were identified to classify the donors and we were able to correctly identify up to 90% of at-risk donors. As a result of this thesis, a screening tool was developed and provided to AVIS. The posterior analysis of the Bayesian models reveals a strong correlation between metabolic syndrome and lifestyle-related variables such as BMI, physical activity levels, and smoking habits. The final classification results are very promising and will allow AVIS to more effectively identify at-risk donors before the onset of metabolic syndrome, optimizing resources and improving donors’ health.File | Dimensione | Formato | |
---|---|---|---|
2024_07_Arrigoni_Tesi.pdf
accessibile in internet per tutti
Descrizione: Testo della tesi
Dimensione
2.7 MB
Formato
Adobe PDF
|
2.7 MB | Adobe PDF | Visualizza/Apri |
2024_07_Arrigoni_Executive_Summary.pdf
accessibile in internet per tutti
Descrizione: Executive summary
Dimensione
592.54 kB
Formato
Adobe PDF
|
592.54 kB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/223374