Fasilkom UI Doctoral Candidate Proposes a FastConvolutional Method to Detect Sleeping Stages in People with Autism

Fakultas Ilmu Komputer Universitas Indonesia > E-News > Fasilkom UI Doctoral Candidate Proposes a FastConvolutional Method to Detect Sleeping Stages in People with Autism


Doctoral Program of Computer Science UI granted a doctoral degree to Ms. Intan Nurma Yulita, who completed her doctoral dissertation entitled “Metode Fast Convolutional untuk mendeteksi Fase Tidur Penyandang Autisme/ FastConvolutional Method to Detect Sleeping Stages in People with Autism.” Her research offers a new method to overcome a patterned sleeping disorder in children with autism, that is by classifiying their sleeping stages. The method is called FastConvolutional.

One of human critical necessities is to sleep. If the number of hours to rest and sleep is not sufficient; it will cause devaluation of abilities and consentrations, loss of strength, high blood pressures and irritability. This condition when happens to children with autism will worsen the children deviant behaviors. Intan Nurma Yulita analysized that the condition of sleep deprivation in children with autism is concerning, due children with autism have weakness in social functions therefore, it is difficult for them to communicate their circumstances to others. The situation like this will compel parents or guardians to be able to identify directly their sleeping problems. In general, this issue concurrents among teenagers and adults that it needs to be addressed and found a solution.

One way to recognize sleeping disorder within children with austism is through Polysomnography Examination. The examination is crucial as the first step to determine type of sleeping disorder and therapy solution by recording patient’s sleep quality while patient is asleep. After recording is taken, doctor evaluates it. Analysis is undertaken by scoring every segment of the recording. The score sets to be a golden standard in carrying out sleep analysis. The sleep cycle of a patient is obtained through scores from every segment of the recording. The doctor examination towards the recording will take up days to analyze. Thus, one way offered to undergo the process efficiently, an automatic application needs to be built to deliver scores from the sleep classification process step.

Intan Nurma Yulita encountered a number of challenges in sleep classification through polysomnography from patient with autism, some of them were challenge in constructing sets of data of people with autism and possible errors that could occur during recording process. The other challenge was to extract relevant features so that sleep stages would be recognized; sleep stage components, and their ability to form a suitable model matched with characteristics and patient with autism dataset. Nevertheless, the research has found a model that recognizes stages of sleep through polysomnography from patient with autism. The main contribution of this research is that the discovery of FastConvolutional method. According to the experiment that was performed, FastConvolutional is a promising classifier in classifying sleep stages. The other new discovery is this research found distinctive features from every stage of sleep in people with autism. She is the 70th doctoral graduate of computer science UI; Ms. Intan Nurma Yulita accepted a very satisfactory predicate. Her dissertation oral defense was performed in Fasilkom Auditorium UI Depok, headed by Mirna Adriani, Ph.D. and promoted by Dr. Eng. Mohamad Ivan Fanany and Prof. Dr. Ir. Aniati Murni Arymurthy on 16 July 2018.