The need for PHM in the field of medical devices and instrumentation is very urgent. When a notebook computer or a transmission cable fails, there is a monetary cost to the user or the manufacturing company. In contrast, medical devices failure can directly lead to fatality. Many of us in good health tend to forget about the importance of medical device reliability and robustness. The breakdown of virtually any medical instrument – be it a cardiac pacemaker, an EKG machine, a dialyzer, a bypass pump, a glucose monitor, or an artificial heart valve – can have deadly consequences for the patients who rely on them. Deadly consequences lead to grief, lawsuits, and severe monetary losses for families, physicians and biomedical companies. Though PHM has been widely used in the area of computer engineering, its applications to medical instrumentation have been largely overlooked. This is an important area of research with the potential to have enormous practical applications and return on investment in the near future.
There are countless examples of these consequences in recent history. In 2005, Guidant Corporation initiated a voluntary recall involving 78,000 of its pacemakers that were implanted worldwide. The reason for the recall was a gradual degradation in the hermetic sealing material in the device causing the electronic circuit to fail. Guidant has also recalled over 50000 implantable defibrillators (half with electronic problems and half with computer memory errors), which have already caused several deaths since 2005. This has resulted in a $296 million USD fee imposed on Guidant by the United States Department of Justice. The case still has not been settled. In 2009, Medtronic sent out notifications to doctors regarding two new models of pacemakers that will likely fail because of faster-than-expected battery depletion rates. Tens of thousands of these pacemakers have already been implanted in patients since their production in 1997.
Medical device failures are not restricted to electronic cardiac implants. Unforeseen fractures in Bjork-Shiley heart valves produced in the eighties have been linked to hundreds of premature deaths. Baxter recalled some 34,000 drug infusion pumps losing some $15 million USD. These pumps are used during surgeries to deliver critical medications and anesthesia to the patient’s bloodstream but were ceasing to function at random intervals. Along a similar a line, a case study from the University of Florida Department of Emergency Medicine describes an automated drug dispensing unit (ADU) which unexpectedly malfunctioned during a life-or-death emergency. After a critically injured patient was rushed into the emergency room, the ADU displayed an uninformative error message (“Printer not available”) and the keyboard was unresponsive. The medication bins of the ADU were not labeled, so the patient nearly died while hospital staff ran to the nearest pharmacy to obtain the required medications manually.
Obviously these are all extremely stressful and costly situations. Without any means of continuously monitoring the system health of medical devices, these situations will continue to occur. Clearly the demand for PHM integration with medical devices is even greater than the demand for general purpose consumer appliances.
Given the complexity and importance of medical instrumentation, the incorporation of PHM techniques into modern devices is critical for the healthcare industry. Prognostics used in the electronics industry can be applied to the healthcare sector in many areas. Many medical device manufacturers have already integrated rudimentary prognostics into their product development practices, allowing them to better predict the operating reliability of both medical instruments and the health of human body.
Although PHM technology is widely used in electronic devices, application in the healthcare sector may be quite different from most home appliances that we use on a daily basis due to more stringent requirements necessary for life-critical support. These devices must be manufactured to meet exact standards so that precision and accuracy can be maintained at all times irrespective of the ambient environment and operating conditions. To illustrate such differences, consider an example of audio signal processing by comparing two scenarios: music playback using MP3 player, and thermal therapeutic ultrasound in an acoustic surgery for treating internal bleeding. Consider what happens when either device fails. The former will mean suspension of music, whereas the latter would mean unstoppable bleeding. Clearly much more detail and robustness must go into the design of PHM systems for medical devices. PHM not only takes care of medical device reliability, it also covers all related areas including biosensor data acquisition to processing and storage in electronic patient records (EPRs) that assures maximum operational reliability of the entire healthcare system.
Despite the greater amount of detail required for PHM techniques in electronic medical devices, the same overarching principles can be used. In a hand-held phone, for example, we may choose to monitor a certain number of capacitors for abnormalities or cracks. In a pacemaker we may do something similar, but monitor two or three times as many electrical components. With the phone we may conduct an FMMEA analysis to correlate temperature changes, vibrations, or mechanical damage to electronic failures. In the pacemaker, we may examine different parameters by employing different types of sensors (since temperature changes and rough mechanical impacts are unlikely inside the body). The prognostics and qualifications have to be modified to the specific device, its function, and its environment.
What is an example of a current healthcare project that the CityU PHM Centre is dealing with? Currently we are researching PHM techniques as they apply to wireless telecare networks serving the rural mainland Chinese population. Wireless telecare networks are used in many first-world countries to remotely care for people with special medical needs. It is a way to continuously monitor for emergencies, such as a simple fall of an elderly person in a nursing home or an unexpected rise in glucose levels of a diabetic patient in his or her home. By utilizing telecare technologies, medical personnel can be informed of such emergencies immediately and take action before it is too late. Telecare is also an important potential technology for rural or developing regions where hospitals and clinics may be sparse and inaccessible to large portions of the population. In situations like this, telecare gives physicians and nurses a way to communicate with and diagnose patients remotely. In a country such as China – with over 500 million people living in underdeveloped regions – telecare technology could not only be a massive business opportunity, but also greatly benefit the Chinese government, which recently enacted a $123 billion plan to establish universal healthcare for all Chinese citizens.
What is the role of PHM in wireless telecare networks? The major objective of telecare is to provide assistance to the sick and injured. This involves sensors that detect different events (such as the mechanical shock of a fall). These sensors will feed data into a communications network (such as broadband wireless access or cable TV). The issue is that these networks have to communicate over extremely long distances and in areas where there is no cellular coverage. As with any signal transmission network, these systems will be subject to various forms of performance degradation as well as network outages. Depending on the situation at the patient’s end of the link, a network outage or delay could be deadly. This is where PHM comes into play.
There are a number of environmental factors that affect the availability of virtually all outdoor wireless networks irrespective of types and deployment options. The weakest link of an entire telecare system is of course the wireless network, which could span anything from a few kilometers across a city to thousands of kilometers across an ocean. Along the path of signal propagation, there are a number of potential disrupting factors, such as:
The most prominent issue with wireless transmission is rain induced attenuation and depolarization, which can result in over 20 dB/km loss in severe cases. This effect will be a major problem in the south-east coastal regions of China, which are affected by tropical cyclones from May until September. It should also be considered that other reliability issues for a telecare system may arise internally due to data corruption or system failures.
Most network breakdowns are due to stochastic link failures, where statistical modeling can describe occurrence due to certain events. The prognostic techniques implemented by the CityU PHM Centre rely on information about network data traffic collected by specialized sensors. This data is analyzed in order to ensure maximum reliability and availability by detecting potential and future problems. CityU PHM Centre is also working on different modulation schemes for network optimization. Although QPSK is a robust solution for long ranges, wider spectra may be more appropriate for areas of less rain. We are also considering the optimization problem of hub placement to maximize network reliability.
The implementation of telecare in rural China is a challenge of staggering proportions, involving problems in statistical prognostics, modulation schemes, adaptive power control, and hub placement. CityU PHM Centre is confident that PHM techniques can be implemented to overcome these barriers.
