Sleep represents a third of our life, from birth to death. Sleep allows our body and mind to rest, and breaking its structure may lead to severe physical and nervous damage. Breathing disorders, like apneas, hypopneas or RERA events, alter the recovering feature of sleep by fragmenting its structure. They usually lead to daytime sleepiness, depression, hypertension, cardiovascular disease,... In order to give the most suitable treatment to a patient, the gold standard polysomnography (PSG) is recorded in a hospital setting and the huge amount of data is visually analyzed the day after. The PSG is expensive, time-consuming for the clinicians and unpleasant for the patient. Thus, portable monitoring devices and automatic analysis methods are welcome. Four physiological parameters are required to score the three breathing disorders mentioned above: nasal airflow, oximetry, arousals and respiratory effort markers. While arousals are defined in the EEG traces, the esophageal pressure is the gold standard but invasive measure of effort. Surrogates (signals) exist for both arousal, like PAT, PTT or ECG, and effort markers, like TAM, PTT or FOT. This thesis was dedicated to a novel one, the maxillo-mandibular movements. This signal is not only able to point arousals and effort, but it has also the capability to distinguish sleep from wake as a mandibular actimeter, like the wrist actigraphy. These three features make it worth of interest. At first, the jaw movements signal essence was extracted, automatic methods 1) to point arousals, 2) to indicate periodic patterns like respiratory effort or salvo of sleep events, 3) to detect and classify apneas, hypopneas and RERA and 4) to separate sleep from wake were developed and evaluated. Then, the sleep apneas/hypopneas and the sleep/wake detectors were then improved by adding the oximetry in a first step. Finally, the nasal airflow brought its potential in both detection and classification of breathing disorders, especially to overcome the inherent classification problem between apneas and hypopneas since the jaw movements sensor is an effort sensor. All the methods developed in this thesis were applied to a huge database of 150 consecutive recordings at the Sleep Laboratory of the University of Liege for sleep apneas and hypopneas detection assessment. Moreover, an APAP device, that applies a regulated pressure throughout a nasal mask to prevent from upper airways collapse, was designed using only features computed from jaw movements in real-time, and showed similar results to the widely tested iS20i from BREAS. In conclusion, the maxillo-mandibular movements signal does bring usefull information about respiratory effort and arousals, and coupled with the nasal flow and oximetry signal provides an accurate detection and classification of sleep apneas, hypopneas and RERA. Besides, this jaw actimeter and its ad-hoc algorithm allows to distinguish sleep from wake. All in all, the jaw movements signal is a very valuable and a unique physiological signal for home sleep studies.