| Objective: In the study of chronobiology,the collection of data on activity,autonomous behavior,physiological signal(body temperature,blood pressure,heart rate,blood sugar,blood oxygen,EEG,ECG and EMG data,breath,exercise,sleep,neural signal,etc.)is the basis for studying the temporal structure and patterns of biological variables.The accuracy of the time series data to biological variables directly determines the correctness of the biorhythm structure that follow.For this reason,there is an urgent requirement to have certain instruments,which are capable of recording and analyzing animal and human rhythm signal continuously and automatically for a long period.But traditional animal rhythm monitoring devices and analysis methods,such as running wheel cages,Actogram and Periodogram,have been unable to meet the above experimental requirements.With the development of computer,software,VLSI,embedded system and wireless communications technology,physiological signal telemetry system has been enormous studied and applied for the animal experiments to obtain the direct measurement of signal source,high sample rate,and actual record from free-moving animals.These general-purpose physiological signal telemetry devices,which generally use dedicated frequency shift keying(FSK)communication,are highly reliable and have fast transmission rates,but require dedicated transmitting and receiving devices,have complex systems,are very expensive,are difficult to deploy,and critically do not fully meet some of the technical specifications required for long-term rhythm monitoring and have corresponding mathematical analysis functions for rhythm data.Therefore,the study of a dedicated signal telemetry system for biological rhythm monitoring based on Io T communication can,on the one hand,make up for the deficiencies of traditional rhythm experimental devices and use the high-resolution and high-precision experimental data obtained on this basis,which can broaden the current rhythm research methods through new ways of presentation,management,calculation and analysis,and is expected to promote new discoveries in the field of chronobiology;on the other hand,using a universal Io T Internet communication,without the need for special data receivers and signal networks,the system not only enables lightweight and rapid deployment,which greatly saves experimental costs,but also facilitates cloud storage and cloud computing of experimental data with the support of the Internet of Things,which is in line with the development direction of data science.This study will aim to solve the following five problems: 1)build an Io T-based rhythm signal telemetry system,including embedded/wearable signal collectors,a general-purpose cell phone and its app as data receivers,etc.,to verify the efficacy and feasibility of Io T and its related software and hardware systems in animal signal telemetry;2)design a highprecision and miniaturized collector that can sample activity and physiological data simultaneously,and through the ultra-low power consumption control of the embedded system,to meet the requirements of animal rhythm experiments with long time continuous monitoring;3)to analyze and extract the recent rhythm signals from the Io T acquisition data through group comparison experiments of rats under different light and dark cycle conditions,and to verify the effectiveness of whether the data can be well used for chronobiological research;4)to design the circadian rhythm algorithm for rat telemetry data,including conversion,calibration,correction,fitting and smoothing,to improve data accuracy and meet the data visualization requirements of the experimental process;further to adapt and improve the traditional Cosinor model used for rhythm analysis through pre-processing algorithms such as Fourier transform,to avoid the impact of accidental data loss and error problems of non-industrial Io T transmission on the results of rhythm calculation;5)use the system to sample the high-resolution activity data,and design mathematical models for automatic identification of autonomous behavioral characteristics of rats to explore its further application value in behavioral rhythm analysis.Methods: According to the performance comparison of several common Io T protocols,as well as the technical index and application scenario requirements of animal rhythm experiments,ultra-low-power Bluetooth(BLE)was selected as the communication standard of this telemetry system,thus determining the main components of the rhythm telemetry system as follows: including the ultra-low-power signal collector based on BLE,the general-purpose mobile phone/tablet and its app as the data receiver,the data analysis computer and its rhythm algorithm model,and cloud storage database.The overall structure and experimental scheme of the system was designed for simultaneous monitoring of activity and physiological signal rhythms with rats as the specific experimental object: the signal collector installed on the head of rats collects their activity signals,physiological signals represented by skin temperature,and room temperature signals as temperature compensation;all the collected data are transmitted to the paired Android cell phone or tablet PC via BLE in real time;the cell phone app initially organized and classified the received data,and then stored them in the phone memory as cache files in the form of database,and then uploaded the data to the cloud storage server through Wi-Fi;the computer analyzed the rhythm and finally got the result report according to the designed mathematical model program by directly accessing the cloud data or local files.Through the signal processing,rhythm calculation and autonomous behavioral action recognition of activity and skin temperature of grouped rats under different light and dark cycle conditions,the feasibility and practicality of this biological rhythm telemetry system were verified,and the correctness of the designed mathematical model of circadian rhythms was also proved.From this,the detailed structure and circuit design of the signal acquisitior was first carried out.According to different animal’s needs,two kinds of collector motherboards were designed,A and B,which of shells both were made of biocompatible materials.The fully wrapped A type collector,with the size of 25*18*8.8mm,could be used with the electromagnetic wave wireless energy charging module for uninterrupted monitoring of larger animals such as rabbits and dogs.The semi-wrapped B type,with the diameter size of 20 mm of circular shape,could be exchanged for the battery method for rat-like animals.In order to reduce the size and power consumption as much as possible while guaranteeing the function and measurement accuracy,the So C chip integrating the ultra-low power embedded processor and BLE core was used as the core microprocessor of the collector,and the structure of each unit was designed based on this;the micro-power MEMS three-axis acceleration sensor ADXL362 was used as the animal motion detection,room temperature detection and system energy consumption control core;the ultra-low-power FRAM was used as the memory data cache;the motherboard was reserved to connect EEG,ECG and other physiological signal sensor boards with IIC or SPI bus by multi-functional generalpurpose I/O interface.Two temperature sensor boards were designed.One used a high-precision current source with NTC as the temperature sensor,and the other one directly used a digital temperature sensor Si7051.Both could be connected to the motherboard with a soft circuit row(FPC),corresponding to two skin temperature detection schemes for A and B collectors,respectively.The system circuit and operating mode were processed and optimized for ultra-low power consumption,and the total average current and power consumption of the whole device were evaluated by calculation method.Combined with the actual engineering experience,the theoretical life cycle was determined according to the selected battery capacity and size,and in the actual circuit test and operation test,one CR1632 battery could support the B-type collector and work continuously for about 21 days at least under the experimental conditions,and the A-type collector was also tested for remote wireless charging using 915 M electromagnetic waves;the design specifications were well met,taking into account the premise of a non-mass-produced prototype.Then the Firmware and Android App software of the collector based on ARM cortex-M0 were designed,containing the normal workflow and fault handling process and functional modules such as Bluetooth connection establishment and interruption,sensor signal processing,signal transmission,data storage,hibernation and energy saving,humanmachine interaction mode;the data dictionary and communication protocol of the whole software system are defined,and the data formats of sampled real-time data,historical data,system time,system status,operation commands,alarm information,and the active and passive interaction methods in Firmware and Android App were determined;the local database and remote database design were carried out for SQLite used in the App and My SQL in the cloud platform;the data exchange between local cache and remote cloud storage was realized by using socket communication,and the test effect on Sina cloud platform was good;through joint debugging,the entire workflow of signal collector,App and cloud for telemetry signal data acquisition,transmission and storage was finally realized,and the system performed stably and reliably in performance test and actual operation.According to the experimental scheme and procedure,an animal rhythm experimental platform was built: rats with collectors mounted on their heads were divided into two groups,which freely moved,watered and fed in the independent cage under the light-dark cycle of complete darkness(DD)and natural day-night(LD 12:12),respectively.The signal sampling was carried out continuously for one month at a rate of 6 Hz.Results: For the skin temperature,which represented physiological data,firstly,the conversion,calibration,correction and fitting of the system voltage and temperature measurements were carried out through the noise and error analysis of the experimental data,and the temperature detection accuracy reached ±0.1℃ in the interval from 30 to 45℃ with the support of oversampling technology;then,according to the characteristics of the temperature curve under high-frequency sampling,the change trend was expressed smoothly using LOWESS;finally Using the intuitive display of the fitted curve,the rhythm visualization of the experimental process data was realized,which can be directly used to predict the experimental results and intermediate analysis.For the activity,according to the characteristics of the MEMS acceleration sensor output data,combined with the characteristics that the movement frequency of the rat’s head activity will not exceed 8 consecutive movements per second,through the comparison of several algorithms,it was determined to use the FFT low-pass filtering with a cut-off frequency of 0.125 to remove the noise in the signal,so that the measurement results can better reflect the real value.After the above processing,the initialized time series data were obtained,which was ready for the accurate analysis of temperature and activity rhythms in the later stage.Then the algorithm model of activity was designed by combining Fourier transform,Cosinor and multiple-Cosinor method.The process was: firstly,the relative acceleration of three axials were vector synthesized by minute to obtain the spatial motion amplitude;the results were then filtered by FFT and then accurately solved by using CAT toolkit based on R language and R Studio,using Consinor and multiple-Consinor method and adjusting their least squares parameters by matching to obtain the circadian;for temperature data,the rhythm analysis was calculated directly by using CAT package after filtering.The results showed that the rats’ circadian rhythms matched perfectly with their light and dark cycles: the rats’ body temperature under the L/D cycle had the most significant rhythm at 22 h and a near-5-day rhythm,and the activity had the most significant rhythm at 24.9 h and a near-halfday rhythm at 8 h.In the rats under the DD cycle,the rats were most affected by the endogenous cycle of 24-h from long-term life,which the original regulatory mechanism was affected by complete darkness,gradually causing the rhythm to shift outward,and the analysis results verified the existence of a 25 h perihelion rhythm for body temperature;the activity perihelion rhythm was strongest at 25.3 h,followed by about23 h.The P values of all the above data were close to 0,which proved the validity of the statistical results.Conclusions: In the present study,an BLE Internet of Things-based rat rhythm signal telemetry system and a circadian rhythm algorithm model based on its dataset were constructed,which can be used to simultaneously acquire the activity and physiological signals of experimental animals with good stability,high accuracy and resolution,and meet the requirements of unconstrained animal rhythm experiments with long-time continuous monitoring,demonstrating that the Io T-based rhythm signal telemetry system scheme well meets the needs of acquisition,transmission,storage and analysis of animal rhythm signals,and had good practical application value and in-depth research value.Further,the system test and animal experiment results showed that about 500,000 motion and temperature records generated by each rat per day in the experiment can be basically and reliably transmitted to the cell phone as the receiver with an overall data loss rate much lower than 0.5%;combined with the time-stamped label of each record,the algorithm model showed strong robustness to data loss and errors caused by BLE communication failures,and errors caused by incomplete transmission data were avoided.Thus,it is demonstrated that the Io T-based signal telemetry system and the circadian rhythm algorithm model we studied provided a lightweight experimental platform for long-term circadian rhythm monitoring in free-ranging rodents and are in line with the trends of visualization,networking,cloud computing,and big data for animal experimental data.Finally,we proposed a method to assess the activity level of animals by calculating the average speed and distance of movement per unit time based on the mobility data,which layed the foundation for later calculations such as calorie consumption.In addition,the acceleration signal characteristics of autonomous behaviors such as drinking,eating and grooming of rats were studied by simultaneous video comparison,and a decision model for action feature classification extraction was established: using K-means clustering,four acceleration clustering centers were obtained by convergence,categorized by Euclidean distance to the minimum of the four clustering points,and finally,according to the categorization contained in each acceleration data set of the action ratio,the autonomous behavior of its representation was automatically identified.The results provided reference data for the classification of characteristic actions based on acceleration signals and layed a research foundation for the analysis of animal autonomous behavioral rhythms. |
PDF Full Text Request