基于光纤传感器提取心率呼吸体动在离床数据

基于光纤传感器采集的光功序列数据，预测心率、呼吸、体动和在离床状态，用于判断个人身体健康状况，是否存在潜在疾病，并做出预警。1.预处理光功序列数据，将序列分成多个固定时长的数据片段；2.将固定时长的数据片段输入滤波器处理之后，分离出心率滤波值、呼吸滤波值和体动滤波值；3.分别对心率滤波值、呼吸滤波值和体动滤波值进行小波变换，获取相应时频域对应的幅值，根据幅值大小选择最大幅值对应的频率，与相应权重值加权求和得到初步的心率、呼吸和体动；4.将初步的心率、呼吸、体动及时频域对应的幅值送入神经网络获得最终的心率、呼吸和体动；5.根据呼吸滤波值对应人体呼吸率频段的信号信息，判断在离床状态，即在床状态下，滤波后的信号具有规律波动趋势，会随着人体胸腔舒张产生相应的信号波动；反之滤波后的信号没有波动趋势。

This dataset utilizes optical power sequence data collected via fiber optic sensors to predict heart rate, respiratory rate, body movement, and bed-exit status, aiming to evaluate individual physical health, identify potential illnesses, and trigger early warnings. The detailed processing workflow includes:
1. Preprocess the collected optical power sequence data by splitting it into multiple fixed-duration data segments;
2. Pass the fixed-duration data segments through a filter to isolate the filtered heart rate values, filtered respiratory rate values, and filtered body movement values;
3. Conduct wavelet transform on each of the three filtered value sets to acquire their respective time-frequency domain amplitudes. For each category, select the frequency corresponding to the maximum amplitude, then perform weighted summation with matching weight values to generate preliminary estimates of heart rate, respiratory rate, and body movement;
4. Feed the preliminary estimates of heart rate, respiratory rate, body movement, along with their corresponding time-frequency domain amplitudes, into a neural network to obtain the final predicted heart rate, respiratory rate, and body movement results;
5. Determine the bed-exit status using the signal information within the human respiratory rate frequency band extracted from the filtered respiratory rate values. Specifically, when the subject is in bed, the filtered signal exhibits a regular fluctuating pattern that aligns with the respiratory motions of the human thoracic cavity; conversely, when the subject is not in bed, the filtered signal shows no such fluctuating behavior.

搜集汇总

数据集介绍

特点

该数据集包含通过光纤传感器采集的59931条数据，每日更新，用于预测心率、呼吸、体动和在离床状态，以评估个人健康状况和疾病预警。数据处理方法包括预处理、滤波、小波变换和神经网络分析。