Browsing by Subject "DeepLabCut"

Anxiety disorders are the most prevalent mental disorders in the world and often show comorbidity with sleep disorders. The bidirectionality of sleep disturbances and anxiety disorders is a hot topic in research and the understanding of the underlying mechanisms is crucial for the development of successful methods of intervention. Studies on the effect of sleep disturbances on anxiety-like behaviour in rodents, however, are limited and have been so far inconclusive. One factor behind this could relate to the limitations that exist when studying behaviour, hindering the replicability of results. Recently, novel behavioural approaches based on deep learning have been introduced, aiming to automate behavioural tracking, thus minimising the subjectivity of manual scoring that is often required for the assessment of more complex behaviours. The aim of this thesis was to use one of the recently developed deep learning tools, namely DeepLabCut (DLC), as well as a DLC data analysis tool, DeepOF, to examine whether chronic (14-day) sleep fragmentation (SF) affects anxiety-like behaviour in female mice and if one week of recovery period affects this phenotype. For the assessment of behaviour, a battery of behavioural tests was conducted immediately after SF and repeated after a week of recovery. The behavioural test data was then analysed using conventional methods, as well as using DLC and DeepOF for supervised analysis of more complex behaviours. These results show that SF does not increase anxiety-like behaviour in female mice, in both conventional and deep learning measurements of behaviour, but may affect locomotor activity, results which contradict previous findings. Despite this, DLC and DeepOF were found to be reliable and valuable tools for the identification and comparison of complex behaviours in mice, overcoming the oversimplified nature of the conventional behavioural tests.