ALT + + Schriftgröße anpassen
ALT + / Kontrast anpassen
ALT + M Hauptnavigation
ALT + Y Socials
ALT + W Studiengang wählen
ALT + K Homenavigation
ALT + G Bildwechsel
ALT + S Übersicht
ALT + P Funktionsleiste
ALT + O Suche
ALT + N Linke Navigation
ALT + C Inhalt
ALT + Q Quicklinks
ESC Alles zurücksetzen
X
A - keyboard accessible X
A
T

Theses

Men, women, aging and sleep apnea. Sleep apnea prediction from actigraphy and polysomnography data from women and men by using machine learning techniques

Objective 

The aim of this work is to investigate whether there are differences in the detection of obstructive sleep apnea between male and female patients. The age of the patients will also be taken into account to check whether the symptoms become more severe with age or whether this is due to other factors (overweight, etc.). Therefore, the main objective is to find the best features from the signals collected from patients and develop machine learning architectures to obtain the best results in the prediction of sleep apnea. Either by distinguishing between men and women or by using data from all patients in general. Ultimately, the study should present the differences in predicting apnea by distinguishing between men and women, as well as by age ranges. 

Description 

This work is interesting since detection in female patients is more difficult to detect than in males due in large part to the fact that obstructive apnea episodes are shorter, so they become almost imperceptible. A clear example is that while snoring, gasping, and witnessed apneas are classic symptoms exhibited often by men the signs differ slightly for women. Polysomnography data from the National Sleep Research Resource (NSRR) can be used for this work and research should be done on machine learning models that generate good results in predicting sleep apnea in women. A study on the performance of the model on male and female data should also be included in this work. 

Resources: 

NSRR :  

Papers of interest:  

 

Novel approach of insomnia prediction from actigraphy and polysomnography data by using machine learning techniques

Objective 

Given the impossibility of having a specific test to detect insomnia, apart from examinations by the doctor, questionnaires... Typically, polysomnography data has not been adequate to detect insomnia, however, this is changing with the advent of artificial intelligence. As a consequence of this, in this work the objective is to investigate machine learning techniques for the detection of insomnia. For this it will be necessary to investigate possible datasets with polysomnography or actigraphy data to study which are the best signals to be used with the algorithms for the detection of insomnia. The goal would be to use as few signals as possible. 

Description 

The Sleep Research Resource (NSRR) can be used to search for datasets. Once the dataset to be used has been selected, the input of the models must be defined, in addition to the selection of the architecture that can work best. The ultimate goal would be to train machine learning models that will yield the best possible results and be a future aid to clinicians. Therefore, models that are explainable will be necessary. For the detection of insomnia, one can classify the sleep phases or search for arousals. Therefore, a clear objective of this work would be to search for the best features for the detection of insomnia. 

Resources: 

NSRR :  

Papers o interest:  

 

 

Does mindfulness or exercise affect to get a good rest?

Objective 

The student aims at finding out a correlation between mindfulness practises and exercise and how it can affect positively to have a better rest. 

Description 

The idea is that the student can be the subject but also the person who carries out the study. A device to measure the quality of sleep and other physiological measures can be used. The student thinks about ways to register the information, for example, the student can do mindfulness before sleeping for several days and keeps their feeling in a diary. Then, the student can compare the results with those days when mindfulness was not practised. An experiment plan should be done. The idea is that the student can give format to the data and also include behavioural variables into the datasets. By doing this, the student would have their own datasets based on the data modelling chosen to make their analysis (over data science techniques) apart from those got by the device. A smartwatch could provide information about the physical activity and also information about the sleep.  

Resources:

Collecting and distributing clinical data using Blockchain technologies

Objective:  

To Enable automated aggregation, replication, and distribution of clinical data among researchers and practitioners with greater traceability and controlled provenance tracking, taking advantage of blockchain technology. 

Description: 

Clinical trial management generates large amounts of data, requiring healthcare administrators to maintain reliable, consistent records for peer review and regulatory compliance. Blockchain tools, in conjunction with electronic data capture (EDC), can help improve clinical data regulation in this sense. Therefore, the idea is to provide some insight of the ways to implement this kind of systems. 

Resources: 

https://onlinelibrary.wiley.com/doi/10.1002/9781119675525.ch13 

https://www.jmir.org/2021/8/e17475/