Prototype app to interpret urinalysis dipstick
I conceived of using a mobile phone to interpret urinalysis dipsticks while doing an away rotation in Ghana during medical school. It occured to me that there is significant inter-observer variability in performing and interpreting urinalysis dipstick results, which greatly limits the utility of the test. Using a smartphone camera to interpret the strip could standardize interpretation (improving accuracy) and provide a mechanism for documenting the observed results. Although resources are limited, smartphones equipped with cameras and urinalysis dipsticks are widely available even in low and middle income countries. In developing and developed countries alike, there are also significant errors in interpreting the results of urine dipstick tests, particularly when the results of each component are not considered together. Therefore, a smartphone app that used the phones camera for acquistion/interpretation of urine dipstick results could also be useful in clinics and EDs worldwide. An app would also potentially be useful for patients to test themselves or thier loved ones.
First some context, urinalysis test strips look like this:
They are interpreted visually by holding this up to the bottle for up to two minutes.
Details of the chemistry and reactions employed can be found here.
Immediately, several challenges are evident:
- different samples require waiting different amounts of time before interpretation (see the schema below for details); reading too soon or too late can cause errors (under-development or over-development)
- aligning the sample (which is covered in urine) to the guide (on the side of the bottle) is necessary to interpret the results; this can be messy and it is easy to misread ('frameshift error')
- there can be significant subjectivity in interpreting the results; for example 8% of men and 0.5% of women are color blind (misreading)
- interpretation is sensitive to lighting conditions (misreading)
- the results must be promptly written down to document them before the strip changes (over-development or transcription error)
- understanding the test results requires the use to combine the results of several test components and interpret the results holistically (e.g. the presence of both leukocytes and nitrites is highly suggestive of UTI)
| Test | Interpretation Time | Result Interpretation |
|---|---|---|
| Leukocytes | 120 seconds | white == negative, purple == positive |
| Nitrite | 60 seconds | white == negative, light pink == positive |
| Urobilinogen | 60 seconds | light pink to dark pink <> 0.2 to 8 mg/ml |
| Protein | 60 seconds | light green == trace, green to dark green <> 1+ to 4+ protein |
| pH | 60 seconds | orange to green <> 5.0 to 8.0 |
| Blood | 60 seconds | yellow w/ green spots to green <> trace to 3+ blood |
| Specific Gravity | 45 seconds | dark green to dark yellow <> 1.000 to 1.030 |
| Ketone | 40 seconds | light pink to dark pink <> trace to 3+ ketones |
| Bilirubin | 30 seconds | white to grey/pink <> trace to 3+ |
| Glucose | 30 seconds | blue to green to brown <> 0 to 2000 mg/ml glucose |
And they are interpreted using a schema such as this:
The app uses augmented reality to overlay instructions and interpretation.
- The app that instructs the user how to use the test strips and has the user hold the phone steady above the test strip.
- With coaching from the app, the user holds the phone aligned above the strip while the phone's light provides standard illumination and the phone's camera acquires images continuously.
- The app interprets each component of the test at the correct time
- The app interprets the overall test results by combining the results from each part of the test and generates a report
- The app securely stores the results and offers the user the option to upload/share them with the medical team (email, SMS, healthkit, EHR integration, etc)
- [CameraX] package for Android
- ARkit used for alignment of the phone
- image acquisition using AVCaptureSession
- MFMailCompose used for email
- HealthKit used for data upload
v0.1.0 this is a work-in-progress
open-source
- Simmerville JA et al, Urinalysis: A Comprehensive Review, Am Fam Physician. 2005 https://www.ncbi.nlm.nih.gov/pubmed/15125171
- Smith GT et al, [Robust dipstick urinalysis using a low-cost, micro-volume slipping manifold and mobile phone] platform(https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4935544/), , Lab Chip 2017
- Wang CS et al, Development of a novel mobile application to detect urine protein for nephrotic syndrome disease monitoring, BMC Med Inform Decis Mak. 2019