Doctors have long known that diabetes affects multiple organs, often leading to complications like diabetic retinopathy and diabetic kidney disease (DKD). It is common for doctors to recommend an eye examination if kidney issues are detected in diabetic patients, and vice versa. This is because uncontrolled diabetes does not isolate its damage to one organ — it spreads its impact, and typically silently.

To improve early diagnosis and accuracy, Chinese researchers have now introduced a cutting-edge artificial intelligence system called DeepDKD, that uses retinal images to detect diabetic kidney disease. Their findings were published in The Lancet Digital Health on April 30 and have already gained significant traction in the medical community.

Traditional Tests Have Limitations

Currently, blood and urine tests, particularly the urine dipstick protein test, are used to detect kidney damage. However, these methods can often fail to distinguish whether the damage is specifically due to diabetes. In fact, according to The Lancet study, only 40% of people with diabetes develop diabetic kidney disease. That leaves a significant margin of error and a massive population that might remain underdiagnosed or misdiagnosed.

Retinal scans, on the other hand, are gaining attention for their potential to reflect broader systemic changes, including kidney damage. The reason? The blood vessels in the retina are remarkably similar to those in the kidneys.

How DeepDKD Works

DeepDKD leverages artificial intelligence and deep learning algorithms to study patterns in retinal fundus images. These are high-resolution images of the interior surface of the eye, including the retina, optic disc, macula, and posterior pole.

The AI was trained using a massive dataset of 734,084 retinal images from 65,406 patients across China, Singapore, Malaysia, Australia, and the UK. This ensures the model’s robustness across ethnic and geographic variations.

In performance tests:

• DeepDKD detected diabetic kidney disease with 77.8% accuracy, significantly outperforming the urine dipstick protein test.

• It also correctly identified cases where kidney disease was not caused by diabetes in 66.3% of cases, another important breakthrough.

This means fewer false positives and improved diagnostic clarity for doctors dealing with complex diabetic complications.

The Significance of Retinal Imaging

As explained by Dr. Hrita Gogate, a Mumbai-based vitreoretinal surgeon, retinal imaging provides visible signs of vascular damage — a symptom that is often mirrored in other organs like the kidneys.

She said, “Urine dipstick can only tell you the level of glucose in the urine. It may be transient. But retinal imaging can show you signs of damage that may be in tandem with other organ damage as well. The attempt using AI may be to detect kidney disease without any invasive procedure.”

The development of mobile fundus cameras could take this innovation even further. The researchers believe that once these affordable retinal scanners become widely available, primary care physicians could use them for on-the-spot screening, making it a reliable, real-time tool for early diagnosis.

Why It Matters for India

India has over 212 million people living with diabetes, making it the global epicentre of the disease. That accounts for over 25% of the global diabetes burden. In such a scenario, a technology like DeepDKD could revolutionise healthcare, especially in rural and underserved regions.

Currently, early detection of kidney disease often fails due to lack of awareness, poor screening infrastructure, and limited accessibility to specialised tests. But if a simple eye scan can alert doctors to the potential onset of kidney disease, it would mean:

• Faster diagnosis

• Early intervention

• Reduced dialysis and transplant needs

• Improved quality of life for patients

China, with 148 million diabetics, also stands to benefit significantly, followed by 42 million in the United States.

Challenges and the Road Ahead

While DeepDKD is a promising tool, clinical adoption is not immediate. There are concerns about:

• Accessibility to fundus cameras in every clinic

• Data privacy concerns around retinal scans

• The need for more real-world testing and standardisation protocols

• Training for healthcare professionals to interpret AI output

Despite these challenges, the future looks bright. As DeepDKD and similar models evolve, the integration of AI in chronic disease management will no longer be a luxury but a necessary healthcare evolution.

Conclusion

The DeepDKD system represents a paradigm shift in the diagnosis and management of diabetic complications. By combining non-invasive retinal imaging with advanced AI analytics, it holds the potential to save millions from irreversible kidney damage. For countries like India and China — grappling with an ever-growing diabetes burden — such innovations could not only improve outcomes but also reduce healthcare costs drastically.

In the coming years, expect to see wider implementation of AI tools like DeepDKD in hospitals, diagnostic centres, and possibly even mobile clinics. It’s a step closer to personalised, precision healthcare for diabetic patients.

The Upcoming IPOs in this week and coming weeks are Srigee DLM, Manoj Jewellers.

The Current active IPO are Wagons Learning, Kenrik Industries, Arunaya Organics.

Start your Stock Market Journey and Apply in IPO by Opening Free Demat Account in Choice Broking FinX.

Join our Trading with CA Abhay Telegram Channel for regular Stock Market Trading and Investment Calls by CA Abhay Varn - SEBI Registered Research Analyst.