There are several research studies we would like to pursue with the HealthMonitor. These include long term monitoring of the elderly. We’re interested in determining which metrics are most valuable for long timer health care, if we can detect acute conditions (such a fall) and be able to notify on such events, and being able to report more detailed and information to their primary care practitioner. To these ends, we need to develop algorithms to detect critical conditions, as well as trending and clustering algorithms to make the wealth of data more manageable and digestible. We are also interested in recording stress indicators (GSR, EMG, heart rate) continuously over long periods of time and seeing how they correlate a number of health parameters such as susceptibility to disease (i.e. cancer, heart disease), sleeping patterns, and productivity. The Berkeley Tricorder Project Pervasive Health Monitoring The Health Monitor 1.8” 1.4” Reza Naima, John Canny University of California, Berkeley Operational Overview Results Introduction The notion of pervasive health monitoring presents us with a paradigm shift from the traditional event-driven model (i.e. go to doctor when sick) to one where we are continuously monitoring a person’s “well-being” through the use of bio-sensors, smart-home technologies, and information networks. This allows us to be more proactive in heath maintenance, as well as allowing the health care provider to make more informed decisions with a greater wealth of accurate data. To these ends, we are working on a very small chest worn device capable of measuring several key health parameters for extended periods of time. The device will be able to store this data locally on flash, detect acute events and send out notification via a Bluetooth network interface using the user’s cell phone, as well as using the Bluetooth interface to synchronize data with the user’s personal computer. We hope this functionality will allow us to gauge long term well being, provide better feedback for ongoing treatments, as well as detect and notify emergency personnel in the event on a serious acute event such as a heart attack. Future Work ECG EMG Pulse Oximeter Audio 3-Axis Accelerometer Low Power Microcontroller (MSP430) TransFlash FAT Filesystem Data Visualization over Bluetooth Sensors Filesystem A C# application has been written to connect to the HealthMonitor and display data in real time. The code has been ported to run on a PocketPC mobile phone for portable monitoring. A FAT16 filesystem has been implemented and optimized to minimize disk I/O. We have been able to achieve maximal write rates of 40k/second. Analog to Digital Conversion The MSP430 is capable of acquiring 12-bit samples of all channels at rates ranging from 8ksps (audio) to 0.1Hz (body temperature) with buffered disk writes and the ability to stream captured data over Bluetooth in real time. General Health Monitoring Effects of Stress Power Consumtion The HealthMonitor has been designed with minimal power usage from the start, however, no measurements have been made on power usage. All components are designed to sleep when not being used, including the microcontroller, and all amplifiers are ultra-low power. Monitor Design Wellness Monitoring Temperature/GSR Next Version Improved analog input stage for ECG and EMG Addition of Pulse Oximeter Addition of USB interface Addition of onboard voltage regulator, status LED Monitoring Chest Sounds Our device will allow us to monitor chest sound such as heart beating, breathing, coughing, and digestion. The audio stage is also configured to feed the Bluetooth IC directly allowing for remote real-time “listening” to the various chest sounds for remote diagnosis by a physician. We also are investigating algorithms for automatic cough detection, asthmatic episode detection. Chest Strap Silicone Stethoscope Rubber Electrodes