Mobile Medical Telepresence

Overview

In conjunction with the Emergency Medicine and Trauma Departments MGH we are building a system that will allow hospital-based physicians to participate remotely in evaluating the condition of patients being transported by ambulance. Potential uses include:

• When an ambulance picks up a patient following a 911 call, the patient is usually transported to an emergency department (ED). This may not always be the best choice: for instance, the patient may be suffering from a communicable disease.
• In some cases, there can be questions about the appropriate destination for a patient. For example, should the patient be taken to a hospital with specialized facilities, to the nearest hospital, to a clinic, or perhaps nowhere.
• For some conditions, e.g., stroke, heart attack, and traumatic injuries; prompt initiation of an appropriate therapy can have a dramatic effect on the outcome. This becomes increasingly important with longer transport times.
• The physical distance and pressure of providing emergent care often prevent detailed and effective data exchange between the teams.

A system that allows real-time audio and visual linkage of the two teams and real-time transmission of pre-hospital data to the hospital could lead to improved outcomes for patients.

Methodology Employed

In metropolitan Boston, there are a number of public carrier wireless cellular channels providing mobile connectivity to the Internet. They are typically optimized for the downstream link, so the upstream bandwidth offered by these channels is less than the advertised rates might lead one to believe. Furthermore, individual channels provide little in the way of quality of service guarantees. These Wireless Wide Area Networks (WWANs) are also dogged by high and variable round trip times, occasional outages, and high losses. However, multiple WWAN providers provide overlapping coverage. Fortunately, preliminary experiments we performed in the Boston area indicate that simultaneous WWAN connections are reasonably uncorrelated. Based on these observations, using multiplexing, or network striping, to aggregate several of these channels to provide virtual channels seems like a plausible solution In fact, we have shown that by taking advantage of service provider diversity and overlapping coverage, we can provide the illusion that a reliable, stable, high-bandwidth channel is available.

The systems consists of a collection of applications (video streaming, audio communications, telemetry transmission, etc), supported by networking middleware. The networking middleware provides network striping capabilities to the applications. Application data is distributed over many network interfaces. The middleware arbitrates between different applications, and optimizes for each application's QoS requirements. The middleware also handles low-level issues related to the network interfaces (e.g., congestion control, disconnections and reconnections). An abstract striping interface is exposed to applications. The set of applications is derived from the needs of our mobile telemedicine project. The video server has, by far, the highest network bandwidth requirements. The video server has been designed to exploit the availability of multiple channels, and to deal with network instability (both in terms of varying bandwidth and varying packet latencies). A separate application handles bi-directional voice communications. Although the voice channel does not need to use network striping to overcome bandwidth limitations, it does take advantage of the multiple network channels to increase the reliability of the audio link. Finally, a third application handles the relatively low-bandwidth telemetry.

Collaborations

This project is a collaboration with MIT CSAIL and Massachusetts General Hospital.