Pervasive Medical Devices: Less Invasive, More Productive
Wireless, robotic, and computer-backed medical devices are interfacing across the landscape, racing to accident scenes, performing surgery, and monitoring long-term care for young and old alike. With the potential for improved healthcare and long-term cost savings, pervasive devices offer healthcare some intriguing new alternatives.
Orthopedic surgeons operate with a hammer, chisel, and saw. In knee replacement surgery, this means cutting where the replacement should go without cutting away too much bone or damaging surrounding sinews and ligaments. This difficult procedure, which depends on a surgeon's accuracy, raises considerable safety issues. That may be about to change.
At the Imperial College, London, Justin P. Cobb, chair of orthopaedic surgery, and his colleagues have developed Acrobot (Active Constraint Robot), a robotic surgical system that constrains partial knee replacement to safe, accurate parameters.
The surgical system consists of a standard PC, a real-time controller board, surgical planning software, and a robotic arm with drills and cutting tools. Imperial College medical-robotics engineers designed the Acrobot prototype, and Imperial College surgeons tested it.
To plan the operation, a surgeon uses the planning software to model the patient on screen using the patient's CT scan. The surgeon then conducts virtual surgery on the damaged knee, putting the replacement in exactly the right place. According to Cobb, this preplanning sets the software boundaries that will enforce the hardware boundaries around what the surgeon may and may not do in live surgery.
During live surgery, the Acrobot arm is manipulated with a force control handle that senses the force and direction of the surgeon's movements. The computer uses this information to compare the surgeon's actions with allowed parameters and applies resistance using brakes to keep the operation within preplanned boundaries (that is, the space in which the surgeon can operate safely and accurately). A color-coded computer screen display also helps guide the surgeon by showing where the drill or cutting tool is within these constraints.
The surgical system knows the robot arm's location at all times based on information from encoders and software. When the surgeon moves the drill to within two millimeters of the safe, accurate boundaries, the robot applies its brakes and the surgeon feels a noticeable resistance. Between two millimeters and the edges of those boundaries, that resistance ramps up exponentially. "You actually have to push to get to the edges," says Cobb.
The purple hardware device suspended above the patient's knee is the Acrobot arm. The knee is open for surgery with rods holding it in position. The surgeon's hands [lower right] operate the force control handle during surgery.Healthcare benefits In most cases, patients have recovered more quickly from robotic than conventional surgery. Patient expenses were lower because they left the hospital sooner.
According to Cobb, surgeons can rely on the technology to operate precisely, with small incisions. Improving surgical accuracy improves patient outcomes—replacements work better and last longer—and could eliminate the need for additional operations.
Another outcome is that the technology makes knee replacement surgery less complicated, making it possible for less skilled surgeons to be very good at it—so it's ultimately less expensive. "When you 'deskill' something, you make it cheaper," says Cobb.
This surgical system will likely face resistance from surgeons or patients who mistrust the technology. However, Cobb reports that their testing showed that Acrobot helped surgeons place artificial knees within two millimeters of planned placement 100 percent of the time. Control group surgeries were this accurate only 40 percent of the time.
While some medical devices such as Acrobot require the patient to come to the technology, many pervasive devices bring the technology to the patient.
Exploring wireless sensor network technology for medical applications, the CodeBlue project ( www.eecs.harvard.edu/~mdw/proj/codeblue) is a collaborative effort of Harvard University, the Boston Medical Center, the Spaulding Rehabilitation Center, the Boston University School of Management, the Johns Hopkins Applied Physics Laboratory, and 10Blade. CodeBlue seeks interest and collaboration from medical researchers, disaster response teams, and the business sector.
CodeBlue's research focuses on wireless sensor networks composed of motes—small devices generally consisting of a processor, memory, low-power radio, antenna, and power supply. The motes incorporate sensing, computing, and radio capabilities in small form factors with low power requirements. This facilitates mobility, allowing motes to gather data such as vital signs continually over long periods within the patient's natural environment. Doctors and nurses can use that data for real-time triage and large-population studies over time, according to Matt Welsh, assistant professor of computer science at Harvard University. Pluto mote. One of the smallest motes is Pluto, a lightweight device that can be worn like a wristwatch. Patients can wear several Pluto motes to monitor such things as heart rate and respiration, explains Welsh. These sensors capture the data and upload it to a PDA, laptop, or PC residing in the patient's home.
The Pluto mote integrates the same components found in larger sensors (microprocessor, memory, battery, and so on) onto a single circuit board, saving size and space. Other space savers include a lightweight rechargeable battery and a surface mount antenna.
The Pluto mote employs an integrated three-axis accelerometer (used in motion studies). "We are working with a group at the Spalding Rehabilitation Hospital, studying patients with Chronic Obstructive Pulmonary Disease (COPD), Parkinson's disease, and stroke," says Welsh. In those applications, researchers look at patient limb movements during various activities. With Parkinson's disease, for example, one complication is tremors, in which the patient is afflicted with unexpected and uncontrollable limb movement. "Some of this is caused by the disease and some is a side effect of the medication that is intended to control it," says Welsh. Pluto motes can help researchers monitor limb movement in order to better predict tremor episodes and adjust medication dosages to head them off.
The Pluto mote, a lightweight sensing device that can be worn like a wristwatch.
Vital Dust mote. The Vital Dust mote-based pulse oximeter uses GPS to track a patient's location en route to the hospital while continuously monitoring heart rate and blood oxygen saturation.
Sensors record vital signs in real time and pass them to a sensor gateway on the ambulance, which forwards the data to the hospital using either a cellular EVDO ( evolution data optimized) connection or an Iridium satellite connection. If the Vital Dust hardware can't connect to the gateway immediately, it can store the data in onboard memory until it can, so that nothing is lost, says Steven Moulton, associate professor of surgery and pediatrics at the Boston University School of Medicine.
In emergency response situations, such pervasive medical devices provide emergency room doctors with information about the number of incoming patients and their vital signs in advance. This helps them better prepare for the patient's arrival.
IBM is integrating existing Bluetooth-enabled sensors and smart phones with its WebSphere technology ( www.ibm.com/websphere) to offer medical-device solutions in pilot tests under the Personal Care Connect Mobile Health Monitoring Solution. According to Kathy Schweda, a pervasive healthcare solutions executive at IBM, the technology and devices take data streams from Bluetooth communications, gather them on a cell phone near the patient, and send that data over an encrypted VPN tunnel to a server. There, the healthcare provider, insurance company, employer, or payer can use the information for such things as managing chronic conditions or monitoring the elderly.
IBM chose Bluetooth over other wireless technologies, such as Wi-Fi, because it consumed less power and was less expensive. Additionally, with its smaller transmission range, Bluetooth keeps data within a five- to six-meter radius. "It is less likely to be hacked," says Schweda.Kidney failure pilot. At the Imperial College in London, IBM is combining its mobile health monitoring with Bluetooth-enabled scales and blood pressure cuffs from A&D Medical to monitor young adults and children with kidney failure. According to Schweda, these devices monitor patient blood pressure and weight to manage their fluid levels. The data, received by an Ericsson smart phone, is transmitted to a backend server, which makes the information available to medical professionals at hand. Nurses can monitor many more children at one time, checking the transmitted data for trends such as weight gain, rising blood pressure, and rising heart rate and alerting a physician when a patient needs attention.
These monitoring tools can also help improve the patients' quality of life. "They can avoid coming in to see their primary care physician or coming in for dialysis earlier," says Schweda.Diabetic monitoring. IBM's Personal Care Connect technology also has a pilot project to monitor diabetic patients. Using Johnson & Johnson's Lifescan glucose meter as the end device gives patients access to a very small, portable device for testing their blood sugar levels that can report the results to healthcare providers. The devices offer another long-term benefit, by providing closer blood sugar monitoring. This helps those monitored correct elevated levels more quickly and delay amputations, according to Schweda. This technology also lets doctors see graphs of the device readings, providing a better idea of how the data varies over time and what it looks like over long periods, says Maria Ebling, research manager of privacy-enabled context technologies at IBM.
Healthcare costs are positively impacted when timely, quality care leads to better outcomes and shorter hospital stays. According to Moulton, this will come through data mining. Mined data, taken via these sensors over time, has a predictive quality; when healthcare providers see similar data with patients in real time, they can act on it immediately, improving their precision.
Most of the technical challenges to the CodeBlue wireless-mote project involve developing sensor networks that can successfully route patient data from multiple sensors while avoiding network collisions. "Three sensors are no problem. Once you get up to six sensors, then the information collides and a lot of it doesn't get through," says Moulton.
Other challenges include adoption, which is impeded by the lack of technical training of many physicians. "As newer, younger people come into the system—people who are comfortable with handheld computers and the Internet, who are knowledgeable of the power of computing—things are going to change," says Moulton. Developers also need to create readable font sizes for the elderly and devices easily used by small children and older adults, says Schweda.
The biggest challenge is getting someone to pay for these technologies. The diabetic-monitoring kits can run upwards of US$1,500 for the blood pressure cuff, scale, glucose monitor, and phone. "The current reimbursement models for the payers don't accommodate these kinds of preventive measures in most cases," says Schweda.