Method and system for detecting abnormal sleep state

The invention relates to the field of detection methods, in particular to an abnormal sleep state detection method and system. The method includes: collecting historical data of heart rate and respiratory rate of a human body in a night sleep period; determining a hypothesis of data probability distribution according to the historical data, and preliminarily determining a form of a probability density function; learning parameters in a probability density function according to the historical data and the hypothesis of the data probability distribution; for new data, according to the learned probability density function, calculating the probability of the data appearing in the historical data; whether the heart rate and the respiration rate are abnormal or not is judged according to the calculated probability, and reminding is conducted; according to the technical scheme, the training speed is high, and the data volume required for training is smaller than that in the prior art; and after the probability model