The App Will See You Now: Smartphones as Diagnostic Devices
The market for health-related smartphone applications (apps) is rapidly expanding, with an estimated 100,000 health-related apps currently available on app marketplaces.1 Some consumer-focused apps allow patients to track diet, exercise, sleep, and weight, like Apple's upcoming Health app that will integrate with other apps to provide a comprehensive snapshot of personal health data. The Mayo Clinic is also currently developing an improved version of their present app that will alert patients when certain parameters (such as blood pressure) are not within the optimal range; the app with then notify the hospital and physician for further care.2 While these advancements may be exciting, the safety, efficacy, accuracy, and risks associated with these apps have not been evaluated in many studies. Also, regulation of the apps by the Food and Drug Administration (FDA) as medical devices has been a challenge for some developers in launching their products.
Efficacy and Reliability
The accuracy and efficacy of apps for diagnostic purposes vary widely based on research evaluating their use by patients and clinicians. Many apps marketed to patients are available for documenting suspicious skin lesions and for the surveillance of skin abnormalities but yield inconsistent or inaccurate results. A 2013 study tested the sensitivity to melanoma categorization by four apps that all claimed or suggested to aid patients in determining if a mole is malignant or benign. Three apps utilized automated algorithms to evaluate the mole and the fourth sent the image to a board-certified dermatologist for review; 188 lesions were evaluated using the apps, in which 60 were melanoma (44 invasive, 16 in situ) and 128 benign. The sensitivities of the apps ranged from 6.8–98.1%, with the fourth application having the highest specificity in melanoma categorization (P<.001 vs. apps one and three; P=.02 vs. app two). However, even the app with the dermatologist review classified 30% of the melanomas in the study as benign.3 A key issue with many apps for patients is that the apps usually lack information on the involvement of medical professionals (if any) in the app development and design.4
Smartphone-based assessments for clinicians have been researched with greater frequency but the findings have been mixed. One study comparing two apps for heart rate monitoring during supraventricular tachycardia (SVT) in pediatric patients found that both had low reliability and accuracy compared to standard continuous ECG monitoring. With both apps, the clinician places the patient's finger over the smartphone camera with the flash engaged. The clinician then measures the heart rate by photoplethysmography, using light to detect changes in the blood via the skin. The first app fared slightly better in accuracy, but failed to provide a measured heart rate in 11 out of 21 attempted measurements and differed from the ECG heart rate by anywhere from 1bpm to -47bpm. The second app failed to provide results in 12 of 17 attempted measurements and differed from the ECG heart rate from +5bmp to -176bpm.5 The use of either app as a replacement to standard ECG monitoring was not recommended by the authors.