Pervasive Biosensor Networks Making health care available everywhere, anytime

A Project Report

Rupesh Gupta 2006EE50413

Vikram Garg 2006EE50415

Pervasive Biosensor Networks Introduction We are rapidly heading towards a future in which computers are pervasive and maybe even invasive. Quite soon we will be wearing devices on our clothes and perhaps even incorporating them into our bodies. These ubiquitous computing devices will create biosensor networks. A biosensor network consists of several networked sensors mounted on or implanted in or around a living being. These sensors continuously collect data about the carrier and the surrounding environment and communicate it to a medical community after a preliminary analysis. This can help alert patients and doctors to changing health conditions before they become dangerous, thus facilitating automated round-the-clock health monitoring. Pervasive healthcare may be defined as the provision of services to patients and medical staff anywhere-anytime in a friendly automated manner. In this study we briefly describe a biosensor network and some of the existing biosensor technologies and their applications. We also discuss some novel ideas and concepts under research or development.

Motivation Apart from the regular health caution alerts for a healthy human, the biosensor networks may be absolutely essential in certain cases. As an example, a sleep apnea patient will usually require an alert caretaker the whole night. This need can be done away with by implanting a biosensor on the patient which will send an alert should the patient suffer cessation of respiration. It may even wake up the user in severe case and suggest a change in sleep position. Another example could be monitoring of hypertensive patients. The condition of a hypertensive patient becomes rather urgent during intensive activity. A physician may request hypertension-related data every night before sleep and early in the morning. All this can be recorded in an automated manner using a biosensor.

Biosensor Networks A biosensor network can be simple: As you step into a room, a sound system begins to play your favorite music. Or they can be very complex: A tiny sensor inside your body monitors your heart rate and sends the information to your physician. The new biosensor micro and nano technologies have rendered these sensors unobtrusive.

Fig1. An illustration of biosensor network

As shown in Fig1. a group of wearable biosensors can be placed at different locations of the human body for continuous monitoring of signals such as ECG, blood pressure, respiration and motion activity (Refer to the appendix for an illustration of wearable biosensors). All these biosensor units are linked to a main console. The console is nothing but a wearable computer to accomplish networking of the biosensors with the medical community. The main functions of the console are, control of the sensing, incorporating intelligence to the sensed data and its preliminary analysis and enabling a secure wireless communication with the medical community. The medical personnel at the other end can respond back to the console with appropriate prescription which is usually displayed on the patient interface. Thus, the console constantly monitors the patient health status and generates alerts and takes actions in significant cases and possible risks.

Biosensor Technologies and Applications There are several pilot studies and projects across the globe to develop novel biosensor network technologies. Here we discuss some of the most promising applications of biosensor networks. Detecting and Diagnosing Diseases Biosensor networks will revolutionize the way doctors detect and diagnose diseases and conditions. Many biosensors are either wearable or are otherwise non-invasive, which could prevent the need for exploratory surgery in particularly difficult cases. Traditionally endoscopy has been performed using long, flexible cords about 9 mm wide which patients find uncomfortable despite sedation during scan. A direct view of the small intestine using optical fibers has remained elusive. Push enteroscopy has had only limited success. Medical device developers have developed camera pills which can be very easily swallowed and once ingested, measure conditions in the gastro-intestinal tract [1]. These endoscope capsules are cheap and eliminate the need for anesthesia.

Fig2. Sayaka endoscope capsule developed by the Japanese RF System Lab The Sayaka endoscope pill (Fig2.) can capture 2 megapixel images at the rate of 30images/sec and transmit to a monitoring device worn by the patient. Eventually, the pill passes through the body and is excreted. On Body Electronics With the advent of functional clothes and body sensors it has become possible to detect diseases at an early stage. The HealthGear system developed by Microsoft Research in 2007 [2] consists of a set of non-invasive physiological sensors wirelessly connected via Bluetooth to a cell phone which stores, transmits and analyzes the physiological data, and presents it to the user in an intelligible

way. It uses a blood oximeter to measure a person's oxygen levels. Other sensors detect and record the patient's pulse. It can even recognize patterns that indicate sleep apnea. Another state-of-the-art example is the 3 years long MyHeart project started in 2004 by the European Union [3]. The developed MyHeart garment uses fabric ECG electrodes and piezoelectric sensors to monitor a person's heart rate and respiratory activity. It also contains an accelerometer, which monitors the wearer's posture and activity level.

Ambient Intelligence Till now sensors are being placed on or inside the body but with today’s growing trend toward intelligence in high end Human Computer Interfaces we have the ability to set up sensor systems in the ambient environment making them truly pervasive. The sensors could be placed on the walls or any suitable location in a room and would be able to measure parameters like temperature of body using simple thermal imaging or other symptoms through facial expression/ gesture recognition. There are several promising machine learning algorithms for symptoms/emotion detection which include techniques such as Bayesian Networks, SVMs and Decision trees which are currently being explored. Until now international databases for facial expression are not linked to symptoms/emotions of the human but the recent progress in Microsoft’s dream project Natal: Milo is being considered as a huge success in this field [4]. In [5] the authors talk about segmenting complex human motion sequences like dance. They have proposed an algorithm (Hierarchical Activity Segmentation) which employs a dynamic hierarchical layered structure to represent the human anatomy, and uses low-level motion parameters to characterize motion in the various layers of this hierarchy, which correspond to different segments of the human body. We propose that the same can be used for detecting gestures like coughing and falling. These can implemented not just using a visual feedback but by using speech, direction of gaze i.e. the same way we are progressing in the field of High end HCI.

Happy Milo

Milo mimics the human motion according to the situation

After asking about home work

Facial expression of a 9 year old boy, asked about his home work Fig3. Microsoft’s dream project Natal: Milo In a nutshell it may even be possible to recognize the state of cardiac arrest by the facial expression of the patient, his/her hand movement. (And probably form the sudden change in the body temperature and the sound he/she makes). This was just an example and a patient might not have some of these signs but with good expression recognition systems along with an extensive classification of the symptoms and the cause, we might be able to achieve an Intelligent Ambient healthcare system.

Conclusion Health monitoring devices have the potential to transform health care by providing doctors and patients with multi-sourced, daily real-time physiological data. As these devices become more pervasive, there is a craving for developing automatic pattern recognition algorithms to model, detect anomalies and ultimately get an understanding of the massive amounts of physiological data. There are certain important characteristics which need to be considered while realizing such devices. These should be lightweight, easy to use and non-intrusive to facilitate at-home monitoring. Apart from this, these real time devices should support wireless communication and easy integration with widely used devices like cell phones. Research is also targeted towards making these devices GPS enabled for effective patient tracking. However all such restrictions may be done away with, if the proposed system of ambient intelligence grows robust over the years.

References [1] McCaffrey, C.; Chevalerias, O.; O'Mathuna, C.; Twomey, K.; Swallowable-Capsule Technology; Pervasive Computing, IEEE; Volume 7, Issue 1, Jan.-March 2008 Page(s):23 - 29 [2] Nuria Oliver, Fernando Flores-Mangas; HealthGear: Automatic Sleep Apnea Detection and Monitoring with a Mobile Phone; JCM 2(2): 1-9 (2007) [3] Joerg Habetha; The MyHeart Project – Fighting Cardiovascular Diseases by Prevention and Early Diagnosis

[4] http://www.youtube.com/watch?v=hw69xCzu5hU&feature=related [5] K Kahol, P Tripathi, S Panchanathan, “Gesture Segmentation in Complex Motion Sequences” IEEE International Conference on Image Processing, 2003 Barcelona, Spain.

Appendix

Fig4. An illustration of a wired wearable health monitoring system

Fig5. A wireless, lightweight ambulatory physiological monitoring system Source: NASA Tech Briefs, Jan01