Fullscreen HTML5Med Res (??MB)HTML4

Medical Diagnosis and Monitoring via Smartphone There have been many recent, exciting developments in medicine which take advantage of the power and ubiquity of the smartphone. This puts powerful diagnostic techniques in the hands of practitioners in the field, or in many cases, patients themselves. Part 1 – Tests which used to take weeks by can now be done in minutes, Dr David Maddison very cheaply. Here are just some of the latest in smartphone-based medical technology. A s the cost of medical care continues to rise, the pressure to reduce costs is mounting. There is also a desire to monitor the patient’s well-being on a continuous basis. One way to reduce cost is to reduce or eliminate the need for patients to visit medical centres for routine tests. If the patient had a suitable testing device, they could perform the test themselves and transmit the results to a medical specialist for evaluation and diagnosis. This would also allow the patient’s condition to be monitored on a regu14 Silicon Chip lar or even continuous basis. Fortunately, today, most people carry with them most of the technology needed to achieve this, probably without realising it. It’s a device which contains a powerful computer and communications system to process and transmit information to a diagnostician. Of course, we’re talking about a smartphone. In some cases, it doesn’t even need any added hardware; the onboard camera, microphone and other sensors such as accelerometers can be used to monitor the patient. Over time, smartphones tend to Australia’s electronics magazine incorporate more and more sensors. There is no reason either why sensors already in separate handheld devices could not be incorporated into a phone. One example is a breath alcohol meter; relatively easily incorporated in a smartphone, it would allow the user to check that they are beneath the legal blood alcohol level before driving. But it could also be useful for many medical purposes. Smartphone medical diagnostic apps can be split into two types: those which use the phone’s inbuilt capabilities, and those which require the addisiliconchip.com.au Fig.1: the Miiskin app is used to document changes in skin lesions, rather than make diagnoses. tion of a peripheral device to perform a function that the smartphone is not intrinsically capable of (possibly in conjunction with other onboard sensors). Examples of the former, described in more detail below, include those which can diagnose an eye condition by imaging the eye directly, while others can read the result of a medical test by sensing the colour that a specially treated paper turns after being exposed to the patient’s blood, saliva etc. Examples in the second category include as a peripheral to detect specific chemical compounds in the patient’s breath which are indicative of disease, and microscope attachments to observe bacteria or genetic markers. Some of the technologies described in this article are already available for use while others are still under development. The technologies described here are only a subset of the hundreds that already exist or are under development. graph your body and compare it with new images taken, say, six months or a year later. Changes in the images automatically flagged information for the specialist to further investigate. While there are still a few of these centres around, they have largely been overtaken by the camera and apps built into smartphones. These apps are now being used to diagnose and document changes in possible skin cancers or pre-cancers. There are at least 26 such apps and while they could be very useful for people living in remote areas, they should not replace regular GP or skin cancer specialist visits. There are also some ethical and other concerns with using these types of apps, which are described in an article at Bioengineering Today: siliconchip. com.au/link/aamf Some of these skin cancer recording and/or diagnostic apps are as follows: Miiskin (https://miiskin.com/) does not attempt to make a diagnosis but is simply a tool for documenting the changes in skin spots over time, as described above (Fig.1). UMSkinCheck (www.uofmhealth. org/patient+and+visitor+guide/myskin-check-app) from the University of Michigan (USA) is another example of a skin cancer app that is used to document changes in possible skin cancer lesions (Fig.2). With SkinVision (www.skinvision. com/), a smartphone is used to take a picture of a suspicious skin spot. It Fig.2: screen grab from the UMSkinCheck app. uses a combination of machine learning and in-house dermatologists for diagnosis. If the software determines that a spot is a high risk for cancer, it is reviewed by a dermatologist (Fig.3). For a scholarly discussion of these apps, see the abstract at: www.ncbi. nlm.nih.gov/pubmed/29292506 DermLite (https://dermlite.com/ products/dermlite-hud) uses a supplemental rechargeable magnifier that uses polarised light in conjunction with the smartphone camera. The device is used to make high-quality Detecting skin cancers Some years ago, a number of skin specialists set up clinics to photosiliconchip.com.au Fig.3: screen grabs of the SkinVision app. It utilises fractal geometry to make its assessment. Australia’s electronics magazine February 2019  15 Medical monitoring apps and devices you may already be using Without realising it, you may already have used apps that could be considered medical in nature. With today’s emphasis on keeping fit, there is a plethora of apps out there for use on smartphones to monitor excercise, heart rate, etc. Others use an app to record what goes in – their food, etc. For example, many people use MyNetDiary (app only) to track their diet, a FitBit (app and hardware) to track exercise and WakeMate (app and peripheral, out of production) to monitor sleep patterns. photographs of suspicious lesions, so that one can share them with one’s dermatologist for review (Fig.4). Incidentally, at least some of these apps are available to download free of charge – if they ask you for your phone number, don’t forget it should be in international format (eg, for Australia, 61401234567). Diabetes monitoring There are numerous apps available to allow diabetic patients to manage their condition by recording what they eat and so on, as well as blood glucose levels. However, most of these require manual entry of test data. We’ll look at some of these in a moment. But there was also an app announcement, in August 2017, from Epic Health (https://epichealth.io/), which is said to be able to use the smartphone camera to check glucose levels. Fig.4: the DermLite HÜD peripheral for photographing suspicious lesions. The app was said to work by having a patient place their finger directly on the smartphone camera and the image is sent to a remote computer for analysis, to determine blood glucose levels based on the patient’s blood flow. However, as of the time of writing, there have been no further announcements on this app and we are somewhat sceptical that this scheme will turn out to be reliable. By contrast, there are Diabetesmonitoring apps which are already in widespread use but they use more invasive techniques, eg, a patch with a tiny needle going into the patient’s body, communicating with the phone via Bluetooth. That’s hardly surprising, given the number of people suffering from (especially) type 2 diabetes – estimated at around 1.2 million in Australia alone and a whopping 422 million worldwide – up by more than 300 million in the past 3 years. Most diabetics monitor their blood sugars manually, using a droplet of blood on dedicated (one-use only) test strip on a blood glucose meter. However, one recently introduced system is the Freestyle Libre from Abbott Laboratories (www.freestylelibre. com.au) – see Figs.5 & 6. It sends data from a tiny needle in a patch worn (usually) on the arm. This automatically transmits readings to a special blood glucose meter which can then transmit the stored data (up to 90 days worth) to an Android or iPhone via another app. This is claimed to be especially useful for parents and caregivers who can monitor blood glucose levels “on the go”. These have been widely promoted recently but it would appear the major reason for lack of acceptance in the diabetic community is, quite simply, their cost, compared to the more traditional blood glucose meters and test strips. (The patch system is not [yet?] subsidised by the Government – ie, on the PBS) whereas use-once test strips are on the PBS) There are yet other diabetes apps which mate with the traditional blood glucose and/or ketone meters that all diabetics know. We’ve seen a couple of these which automatically (or manually) transmit the meter’s readings to a smartphone. This has a possible three-way benefit – (a) it saves the diabetic patient from having to transfer their readings to a diary; (b) the readings can be automatically forwarded on to the patient’s specialist, and (c) some are said to be capable of warning the patient where there are significant changes in readings – especially ketones. The one thing that they don’t do is save the patient from pricking a finger up to several times a day to obtain blood Fig.5: the Freestyle Libre system continuously monitors blood sugar levels via a “patch” worn on the body, which transmits data to the blood glucose meter. It can then send data to a mobile phone via the LibreLinkUp app, as shown in Fig.6, right. 16 Silicon Chip Australia’s electronics magazine siliconchip.com.au Fig.7: the free CRADLE app for iPhone and Android devices, to detect leukocoria from photographs. samples for the meter to analyse (the part that diabetics universally hate!). Diagnosing eye disease A smartphone’s inbuilt camera can be used to diagnose eye problems. While “red eye” is a technical problem that many photographers experience, it’s not indicative of any health problems (it’s caused by the camera’s flash light reflecting off the retina inside the eye. Many cameras “double flash” to make the eye’s iris close down on the first flash and take the photo itself on the second flash). However, so-called “white eye” or leukocoria as seen in photos of both children and adults can be a sign of an underlying condition. Immediate medical attention should be sought if this is noticed. And once again, there is a high probability that in not-too-distant future, a smartphone camera could capture and compare images of the back of the eye to detect the early stages of diabetic neuropathy – a quite common and relatively serious effect of diabetic damage. There is also the possibility of smartphone apps being developed for early diagnoses of cataracts, glaucoma, tunnel vision (retinitis pigmentosa) and other eye disorders. An app called CRADLE assists in detecting early forms of some eye diseases, although it is no substitute for an examination by a medical professional. For more information, see: https://cs.baylor.edu/~hamerly/leuko/ Monitoring Parkinson’s disease symptoms Parkinson’s disease results in tremors, stiffness and slow movements. It is caused by a shortage of the neurotransmitter dopamine and its symptoms can vary widely. Patients are typically assessed by a specialist a few times a year, but these tests are largely subjective and Parkinson’s symptoms are known to vary drastically over time. For improved symptom assessment, it is important to monitor symptoms much more frequently and using more objective criteria. mPower (parkinsonmpower.org/) is a “mobile Parkinson’s observatory for worldwide, evidence-based research”. Fig.8: some sample screens from Sage Bionetworks mPower app for Parkinson’s disease sufferers. siliconchip.com.au Australia’s electronics magazine February 2019  17 Fig.9: this shows how the mPower app works. It is an iOS app designed by Johns Hopkins University and the non-profit organisation Sage Bionetworks (see Figs.8 & 9). The purpose of this app is to participate in a study which allows patients to monitor the progress of their condition on a regular basis, rather than infrequently by medical appointments. Parameters such as gait and balance, spatial memory, finger tapping and walking can be monitored. Data can also be acquired from wearable devices. The patients also have the option to “donate” their data to researchers. Early insights into the disease made by researchers using this data include severity of symptoms as a function of time of day and responses to exercise or treatment. The higher frequency at which data is collected leads to new insights into the disease. Note that only people who live in the USA are currently eligible to participate in this study but there are plans to extend it to other countries. You can view a YouTube video playlist with instructions for the various tests in the HopkinsPD app, the pre- decessor of mPower, at: siliconchip. com.au/link/aamh Using a smartphone for remote diagnosis Sana (http://sana.mit.edu/) is a system intended primarily for use in less developed countries. It provides a smartphone-based platform for communications between a healthcare worker in the field and a remote clinician for remote disease diagnosis and data storage (Fig.10). According to Sana, they use technology to “overcome resource limitations, focusing on analytics to drive evidence-based quality improvement, and an educational program for capacity building to promote locally sustained innovation” in health care. There are various similar projects underway in Haiti, India, Lebanon, Mexico, Philippines, Uganda among other places. See the video titled “Mobile Medical Diagnostics” at: https:// youtu.be/h-Zz5a6ARsQ Smartphone apps for clinical trials and epidemiology Patients undergoing clinical trials Fig.10: the Sana app concept, showing communications between a field worker and clinician and also the intermediate data storage. 18 Silicon Chip with new drugs are often asked to use a smartphone app on a daily basis to self-assess their symptoms, by rating the severity of their condition on a numeric scale. This data is used to determine if there has been an improvement in their condition due to the experimental drug, side effects and so on. Similarly, field health workers (especially in Third World countries) can use smartphone apps to log incidences of disease outbreaks (such as Ebola) into a central database, so their spread can be tracked by authorities. Diagnosis with a smartphone and a passive device In some cases, a smartphone app is paired with a passive device like a skin patch, to perform diagnoses which are not possible with the phone alone We’re referring to these devices as passive since they don’t contain electronics. apps that use external electronics hardware will be described later. Bacticount Bacticount (http://bacticount.com/) is a free, open-source diagnostic sys- Fig.11: the Bacticount methodology. Australia’s electronics magazine siliconchip.com.au The Qualcomm Tricorder X-Prize Fig.12: a series of Biosensors tattooed on pig skin for testing. Top row, from left (a and b) show a glucose biosensor without and with glucose; bottom row, from left (c and d) show a biosensor at pH 7.0 and pH 8.0; top row, last two (e and f) show a sodium sensor in visible light and UV light; bottom row, last two (g and h) show another type of pH sensor under visible light and at pH 8.0 under UV light. Source: MIT Media Lab. tem to identify microbial infections. It is specifically designed to detect urinary tract infections but it can be made to work with other types of infection. It uses a process called smaRTLAMP or real-time loop-mediated isothermal amplification to identify bacteria on specially prepared plates. These fluoresce if specific bacteria are present and the amount of fluorescence can also be used to determine the concentration of the bacteria. Up to 36 samples can be tested at a time (Fig.11). Apart from the phone, the hardware required costs around US$100. The app is limited to the Samsung Galaxy S7 phone due to camera calibration requirements. Bio tattoos Biosensor tattoos have been developed at MIT and Harvard, under the project name “DermalAbyss” or d- abyss. These use the skin itself as an interface to measure parameters such as glucose, pH and sodium levels in the blood. The skin is injected with a biosensor marker which changes colour according to changes in the parameter being measured. The colour change can be accurately measured with the camera of a smartphone. The concept has been tested on pig skin samples in the laboratory; there are currently no plans to bring the project forward to a clinical trial or a product for human use (Figs.12 & 13). Next month: When we started researching this field, we never imagined there were so many apps out there (much more than we could fit in one issue!). So next month, we will conclude this feature with smartphone apps that use additional hardware for diagnosis. SC Similar to other X-Prizes you may have heard of, the Qualcomm Tricorder X-Prize was intended to promote the development of a hand-held medical device, much like the fictional Tricorder from Star Trek. The winning device was to be able to “diagnose and interpret a defined set of 13 health conditions to various degrees, while continuously monitoring five vital health metrics”. (We discussed the Google Lunar X-Prize on page 8 of the November 2018 issue, which was established to encourage private space companies to build a moon lander). No team met all the requirements of the full prize in the required time, but in 2017, the top prize of US$2.6 million was won by the family-lead team Final Frontier Medical Devices and the second prize of US$1 million was won by Dynamical Biomarkers Group. Both devices are mentioned in the second article in this series. DxtER was described as an “artificial intelligence-based engine that learns to diagnose medical conditions by integrating learnings from clinical emergency medicine with data analysis from actual patients. DxtER includes a group of non-invasive sensors that are designed to collect data about vital signs, body chemistry and biological functions. This information is then synthesized in the device’s diagnostic engine to make a quick and accurate assessment”. Dynamical Biomarkers’ device paired “diagnostic algorithms with analytical methodology in a userfriendly device” and was controlled using a smartphone. For more details, see: https://tricorder.xprize .org/ prizes/tricorder Fig.13: some sample colour changes from biosensor tattoos. The specific colours and thus the values being measured can be determined with a smartphone camera and appropriate software. siliconchip.com.au Australia’s electronics magazine February 2019  19