A framework for supporting emergency messages in wireless patient monitoring

Abstract

Patient monitoring is becoming a requirement for offering a better healthcare to an increasing number of patients in nursing homes and hospitals. During the monitoring, vital signs of patients could fluctuate significantly and/or match certain undesirable patterns and therefore “alerts” or emergency messages must be delivered to healthcare professionals. In this paper, we address how patient monitoring, specifically emergency messages, can be supported over wireless ad hoc network formed among patients' devices. The framework describes a wireless patient monitoring system that includes patient monitoring devices, routing protocols and information presentation for vitals signs and parameters. This involves a series of decisions in obtaining and processing vital signs, routing over networks, and delivering to healthcare professionals, who must make suitable medical decisions on patients' healthcare needs. Additionally, several design enhancements to improve the quality and coverage of wireless patient monitoring are presented. The performance results for the proposed ad hoc network based architecture show that reliable message delivery and low monitoring delays can be achieved by using multicast or broadcast-based routing schemes. The proposed monitoring architecture is shown to be scalable and the cognitive load on healthcare professionals is found to be dependent on routing protocols and reliability requirement of emergency messages. The proposed work can be extended to provide personalized healthcare services to people in nursing homes, assisted living, home, and while being mobile.

Introduction

With an increasing cost of healthcare and a growing population of seniors in nursing homes and hospitals worldwide, patient monitoring using wireless technologies is being considered as a solution to both improving the quality of healthcare and reducing the rate of increase for healthcare services [2], [3], [8], [12], [15]. As many of the patients are mobile and therefore could be well served by using wireless networks for monitoring [26], [34], [38], [39], [40]. In general, patient monitoring involves periodic transmission of routine vital signs and transmission of alerting signals when vital signs cross a threshold, patients cross a certain boundary, or device battery drops below a level. There are many challenges in wireless monitoring of patients, including the coverage, reliability and quality of monitoring. The work done in patient monitoring includes home monitoring [19] wireless telemetry system for EEG epilepsy [30], Bluetooth-based system for digitized ECGs [17], a hospital-wide mobile monitoring system [33], mobile telemedicine [1], [10], [31], [32] and, real-time home monitoring of patients [27]. The work on devices and sensors include clothing-embedded transducers for ECG [12], ring-based sensor [35], minimally invasive wireless sensors for health-monitoring [5], [18], and personal health monitors for stress monitoring [13]. More recent work includes support for alerts [14], [20], elderly [23], [29], real-time monitoring [22], and monitoring for disabled [47]. An autonomous intelligent agent for monitoring Alzheimer patients' health care is presented in [7]. A mobile decision making system for using symptoms of abdominal pain in children's emergency is presented in [28]. Some discussion on DSS requirements for emerging applications can be found in [36]. Several ways to improve medical decision making are included in [4], [6], [9], [24].

One of the most difficult challenges in patient monitoring using wireless networks, especially for emergency messages, is the reliability of message delivery. The quality of patient monitoring is also affected by the end-to-end delays or monitoring delays. Additionally, the wireless patient monitoring system should be scalable to support as many patients as possible. Many hospitals and nursing homes are deploying infrastructure-oriented wireless networks, such as wireless LANs, satellites, and cellular and GSM (Global System for Mobile communications) systems, where a fixed infrastructure is utilized to support fixed and mobile patients [41], [45]. The potentially spotty coverage, due to time and location-dependent channel quality and signal attenuation resulting in dead spots, of infrastructure-oriented wireless networks [37] will significantly affect the reliability of emergency message delivery [42], [43], [44]. The resulting unpredictable quality and reliability of patient monitoring could lead to difficulty in achieving continuous patient monitoring and delivery of emergency signals from a patient to healthcare professionals, and eventually, the delayed medical response to patients could result in injury [42]. To support the reliability and monitoring delay requirements of patient monitoring, significant work is necessary in creating wireless network architecture and protocols to support routing and delivery of messages carrying a range of vital signs and healthcare information. To overcome the coverage problems of infrastructure-oriented wireless networks, several patient monitoring devices can form an ad hoc network for transmission of emergency messages carrying digitized vital signs [46]. In this paper, a framework is presented to describe a wireless patient monitoring system consisting of patient monitoring devices, ad hoc networks, devices for healthcare professionals, and, network routing protocols. The novelty in the wireless patient monitoring is the use of ad hoc wireless networks for increasing the reliability of emergency message transmission. The results show that reliable message delivery and low monitoring delays are achieved by using multicast or broadcast-based routing schemes.

Now we discuss how vitals signs are represented and digitized, requirements of emergency signals, and the contributions of this paper to wireless patient monitoring.

The vital signs include ECG, blood pressure, pulse, body core temperature, and oxygen saturation [48]. These vital signs are shown in Fig. 1, where each vital sign is represented as an analog signal along with its nominal value. In addition to the diversity and the number of vital signs, the frequency and representation of relevant healthcare information could affect the management of network traffic, and the achievable reliability and monitoring delay. As shown in Fig. 1, several vital signs are obtained, sampled, and digitized for transmission as network packets. The traffic generated by digitization of vital signs can be compressed [11], however, increased processing and packet delays, and, potential for any “introduced” errors in critical healthcare information must be carefully considered.

The severity of one or more medical conditions reflect in changing values of vital signs such as pulse rate involving bradycardia (less than 60 pulse) and tachycardia (more than 100 pulse), blood pressure, breathing rates, oxygen saturation, and ECG. In an ECG signal, P wave represents the sequential activation (depolarization) of the right and left atria, QRS complex shows right and left ventricular depolarization, and ST-T wave is for ventricular re-polarization. The PR interval is the time interval from onset of atrial depolarization (P wave) to onset of ventricular depolarization (QRS complex) and the QT interval is the duration of ventricular depolarization and re-polarization [49]. Any significant changes in wave pattern may indicate patient-specific cardio-vascular problems such as missing or weaker P wave indicates atrial problems affecting blood flow to the heart and a deformation in the Q wave represents damage to the heart. A large increase in the Q wave with respect to overall QRS indicates myocardial infraction, while inverted T wave indicates ischemia. A depressed ST segment indicates obstructions in the arteries. These conditions could be detected by patient monitoring devices or some devices may perform a simple comparison of current ECG signal with a prior ECG signal to generate an alert.

Our work can utilize either the common and very popular 3-lead ECG system or a more comprehensive but difficult to wear 12-lead ECG system (Fig. 2). The 3-lead system, used in Cardiac Care Units (CCUs) and commonly implemented as Holter monitor, can detect QRS complex and heartbeats as it observes the electrical activities of the inferior and lateral walls of the left ventricles. Thus it can detect bradycardia, tachycardia, sinus arrest, ventricular tachycardia with broad QRS complexes, and supraventricular tachycardia with narrow QRS complexes [25]. The 12-lead ECG provides spatial information on heart's electrical conductivity in three approximately orthogonal directions as it observes electrical activity at lateral, inferior, septal, and anterioral walls. The 3-lead ECG system is more suitable due to its wearability for more patients in nursing homes and availability. In future, as the 12-lead system becomes more integrated in wearable PMD, it can be utilized for a more precise and comprehensive detection of cardiac problems for emergencies.

During the patient monitoring, some vital signs could fluctuate significantly and potentially crossing pre-defined thresholds or matching certain undesirable patterns, resulting in “alerts” or emergency messages. The range and patterns of vitals signs must be highly personalized as different patients may have unique ranges of “normal” vital signs. In some cases, an individual vital sign may not result in an emergency situation, however when combined with few more high-normal or low-normal vital signs, an emergency event may occur requiring the transmission of emergency messages to healthcare professionals. Although very important for patient monitoring, there has been little work on emergency messages in wireless patient monitoring [16]. These emergency messages must be reliably delivered to healthcare professionals with minimal delays and message corruption. Many traditional engineering considerations such as network efficiency [21], traffic, and the number of patients that can be monitored or supported by wireless infrastructure may become secondary. However, these affect the scalability and the cognitive load of healthcare professionals that are involved in patient monitoring.

The primary objective of this paper is to provide a framework to support emergency signal transmission using ad hoc wireless networks, which can be formed among patient-worn devices, including radio-enabled watches and a grid of body sensors. A possible scenario for patient monitoring is shown in Fig. 3, where multiple devices used on a patient can form body area network (BAN) for communications related to different vital signs to healthcare professionals. Both, intra-body BAN and inter-body BAN, involving devices on multiple patients, are ad hoc wireless networks, where topological changes occur with mobility of patients, power transmitted by devices, and the wireless channel characteristics. In practice, patient monitoring could also involve a combination of infrastructure-oriented wireless LANs (with base stations) and ad hoc networks, we focus on ad hoc networks in this paper due to their novelty and potential in supporting emergency messages in patient monitoring.

The framework presented in this paper is designed to support the requirements of patient monitoring, including emergency messages which require very high levels of reliability and low delays. The framework also includes representation for healthcare information and enhancements for network routing in the presence of un-cooperative routers. The novelty in the proposed work involves the use of ad hoc wireless networks for increasing the reliability of emergency message transmission (Section 2). The performance of routing schemes presented in the framework, namely multicast, broadcast, reliable multicast and reliable broadcast for delivery of emergency messages, is also evaluated using analytical modeling techniques (Section 3). The performance results show that reliable message delivery and low monitoring delays can be achieved by using multicast or broadcast-based routing schemes. Using many proposed enhancements, the system is shown to be scalable and the resulting cognitive load on healthcare professionals is found to be reasonable, and relates to the routing protocols used and reliability requirements of patient monitoring.