Photo Credit: Rasi Bhadramani

Noninvasive tests such as retinal imaging and urine dipstick assessments significantly enhanced the effectiveness and efficiency of CKD screening.

A deep learning model that uses both retinal images and urine dipstick data offers a noninvasive way to screen for chronic kidney disease (CKD), according to findings published in JMIR Medical Informatics.

“Evidence suggests that individuals displaying retinal microvascular signs, including retinopathy, arteriolar narrowing, and venular dilatation, exhibit an increased predisposition for CKD, and vice versa,” researchers wrote. “Fundus imaging allows for the convenient assessment of the retinal microvasculature, making it a potential screening modality for incident CKD.”

Because conventional fundus photography alone is limited in its ability to predict CKD, researchers integrated urine dipstick test results with retinal images in a multimodal deep learning (MMDL) model in an attempt to enhance CKD prediction. They also created a retinal image deep learning (RIDL) model and a urine dipstick logistic regression (UDLR) model for comparison.

Each model was trained to predict an estimated glomerular filtration rate (eGFR) of less than 60 mL/min/1.73 m², a key indicator of CKD.

Integration of Retinal Images & Urine Data

In a development set of 65,082 adults and an external validation set of 58,284 adults, the MMDL model outperformed the RIDL model. In the external validation set, the area under the curve was 0.88 for the multimodal model and 0.77 for the retinal image model, according to the study.

The urine data model also outperformed the retinal image model and was comparable to the multimodal model in the external validation set. However, the multimodal model’s performance across all subgroups was superior.

“This suggested that the enhanced performance of eGFR-MMDL was not due to urine data alone, but rather from the synergistic integration of both retinal images and urine data,” study investigators wrote.

The MMDL model performed best in individuals younger than 65 years or those with proteinuria. Consequently, researchers recommended routine blood test screening for adults aged 65 years and older. For other patients, the multimodal model offers significant promise for noninvasive CKD screening, they noted.

“The use of noninvasive tests such as retinal imaging and urine dipstick assessments significantly enhances the effectiveness, accessibility, and efficiency of CKD screening, benefiting both patients and health care providers. For patients, these tests reduce discomfort and anxiety, leading to higher compliance with regular screenings and enabling earlier detection of CKD, which is crucial for preventing progression to end-stage kidney disease.”