Agenda Description of the Process Request Requirement Specification Design Description of the Design Drawing Block Diagram Circuit Schematics Flow Chart Bill of Materials Description of the Implementation Plan Implementation Testing Evaluation
Process - Request The initial written customer request- The parent would like a device that could detect the rhythmic motion their child exhibits during a seizure and set off an audible alarm to alert the night nurse or wake the parents themselves. During the meeting with the customer, the situation was further clarified- Exhibits generalized tonic-clonic (GTC) seizures Seizure detection needs to occur in less than three minutes After researching GTC seizures, two candidate seizure symptoms were identified- Abnormal motion, detectable by an accelerometer Increased heart rate, detectable by a heart rate monitor Dr. Berg, the director of the Strong Epilepsy Center, gave further design considerations- A motion based system could be used to detect multiple types of seizures To detect multiple types of seizures, multiple detection units are needed A library of seizure data could be formed, leading to a robust detection algorithm
Process – Request (Continued) As our project involves not only our customers child, but also patients at Strong Memorial Hospital, further documentation is required- A patient consent form to explain the risks and benefits Health Insurance Portability and Accountability Act (HIPAA) certification to insure patient confidentiality and safety A testing protocol for University of Rochester’s and RIT’s internal review boards (IRB)s Since a robust multi-unit, multi-seizure type detection unit is beyond the scope and time constraints of our original design project, the project was divided into several development phases- I. Develop an accelerometer based worn unit that can wirelessly transmit seizure data. The data will be sent and stored in the base unit. II. Acquire a library of seizure data. Then use this library to develop a seizure detection algorithm for each seizure type. III. Develop a base unit that can implement the seizure detection algorithms and alert a guardian in case of seizure activity.
Process - Requirements 1. Seizure monitoring system during sleep time. 2. Continuous Battery operation up to hours with exchangeable battery. 3. Monitoring system consists of 2 units; base unit and a worn unit. 4. Two units must communicate wirelessly over a distance of 150 feet. 5. No way of harming the wearer. 6. The worn unit will be easily attached/detached by a guardian. 7. The base unit will be capable to store at least 10 seizure events. 8. The base unit will alert the guardian of the seizure activity using an audible alarm on the base unit. 9. Develop a preliminary seizure detection algorithm to process the seizure data. 10. Using the collected seizure data validate the accuracy of the preliminary seizure detection algorithm.
Process - Specifications There may be up to 6 worn units communicating to the base unit to properly monitor seizure activity. A seizure activity will be detected using a 3-axial accelerometer with acceleration values up to 6g. An accelerometer reading will be passed to the microcontroller on the worn unit where it will be checked against an adjustable threshold voltage. ZigBee wireless communication will be used between the worn unit(s) and the base unit. The transfer rate of data between two units will be 250 kbps. The base unit will be capable of writing the seizure data to a USB flash drive with maximum capacity of 1GB in a text file format for later analysis. The base unit will run on a standard wall outlet The worn unit will run on 2 standard AAA batteries Attach/Detach system, for the worn unit, will use a Velcro strap. Size of the worn unit will be approximately 2.5” x 2.5” x 1.5”
Process – Specifications (Continued) 15 Hz max signal (tested) will be sampled at 120 Hz, converted through a 10 bit ADC 3 axis accelerometer, possibly 6 total accelerometers. Therefore bits/s Typical max seizure length = 5 minutes. Therefore 5,184,000 bits < 1 Megabyte Necessary transfer rate from worn to base: 2,880 bits/s ( kb/s) With 6 accelerometers the base unit must handle a transfer rate of 2.11 kb/s Acceleration Info: +/- 6 g’s (tested) Bandwidth maximum = 150 Hz Frequency Resolution is based on sample time, our 5 minutes could produce accuracy up to Hz.
Specifications – Self Test The battery voltage on the worn unit will be constantly monitored when the unit is turned ON. In the event it drops below 2.4V, the “Low Battery LED” on the worn unit will be turned ON. The worn unit will implement a self test each time the unit is turned ON. This self test is used to test the following components and functions of the seizure monitor system: Accelerometer (worn unit) Wireless transmitter (worn unit) Wireless receiver (base unit) To ensure system integrity, the wireless communication system and accelerometer will be tested in addition to the startup test. The “System Operates Properly LED” on the worn unit will be blinking every second unless it is turned OFF during the self test as following:
Specifications – Self Test (Continued) Wireless Communication Test In the event that no signal is transmitted from the worn unit to the base unit during one minute, a test signal will be sent to the base unit. The base unit will detect the test signal and will send it back to the worn unit. In the event that no signal is received back from the base unit, the “System Operates Properly” LED on the worn unit will be turned OFF. In the event that no signal is received by the base unit from the worn unit (including the test signal mentioned above) during one minute, the base unit will trigger the sound alarm to alert the guardian. The “Communication Operates Properly” LED on the base unit will be turned OFF.
Specifications – Self Test (Continued) Accelerometer Test When the worn unit is turned ON, a small, unbalanced DC motor, similar to those used in pagers and cell phones, will be activated for a second. The vibration produced by the DC motor will be detected by the accelerometer and the signal will be transmitted to the base station and written to a file. In the event no signal is transmitted from the accelerometer to the microcontroller in one minute, the unbalanced motor will be turned ON by the microcontroller to “shake” the board. If no signal will be detected from the accelerometer, the “System Operates Properly” LED on the worn unit will be turned OFF and the “System Failure” signal will be sent to the base unit to produce an audible alarm to alert the guardian. A “System Failure” LED on the base unit will be turned ON.
Process – Design Decision Matrix for assessing top 3 candidates for GTC seizure monitoring Scale (1-10 with 10 being the best) Frequently Exhibited Easily Measured/ Ease of Use Magnitude of Variation from Baseline Symptom Occurs Quickly CostSum Body Shakes Breathing Oxygen Saturation level Blood pressure EEG abnormality Body Temperature Heart rate
Process – Design (Continued) Easy to implement using a base and a worn unit? Easy to attach/detach? Accuracy of result? Continuous power for 14 hour operation? Body shakesOKBestOK EEG abnormality detected by multiple electrode EEG recording Worst BestWorst Heart rateBestOKWorstBest Top 3 candidates for GTC Seizure Monitoring
Feasibility: Zigbee communication and accelerometers have been used in similar applications. Signal processing has been accomplished by one of the team members in the past on accelerometer data. Antenna design is a new area for all team members providing some concern. The team will get help from Dr. Venkataraman in designing the antenna. Without a pure way of accessing the battery life there exist some concern in meeting a long life (preliminary calculations have been accomplished). Board layout has been accomplished by two team members in the past, but there is some intrinsic difficulties in any board layout. The design uses devices that have been previously tested on humans, so no foreseen implications from this testing are expected.
Implementation Plan 1. Use the evaluation board that includes the accelerometer unit and the transceiver to develop software code for the microcontroller on the worn unit. 2. Develop software code for the microcontroller on the base unit. 3. Breadboard the worn unit to validate the accuracy of the design. 4. Assemble a prototype worn unit and perform “in house testing”. 5. Make a preliminary printed circuit board layout for the worn unit. 6. Develop a preliminary seizure detection algorithm to process the seizure data. 7. Upon approval from RIT/U of R IRB Review Boards, implement human testing and capture data. 8. Using the collected seizure data validate the accuracy of the preliminary seizure detection algorithm.
Summary Team members gained significant knowledge on seizure types and possible detection techniques. The preliminary design of the seizure monitoring system has been accomplished. As the ZigBee evaluation kit will be used for initial data acquisition, there may be some changes to the design. Upon assembling and testing a prototype, consideration of modifying the design will be made based on the test results and the functionality of the system to meet the desired outcome. Data collection will be orchestrated early in Senior Design II at Strong Memorial Hospital. A low level algorithms for detection will be written and then tested once the seizure data has been collected. Some changes may occur based on the test results.
Acknowledgements: Dr. Phillips Dr. Berg – Director of Center for Epilepsy at Strong Memorial Hospital, Rochester, NY. This material is based upon work supported by the National Science Foundation under Award No. BES FSI International Incorporated Sanyo Our Customer