A creatinine result is not a care plan.
An artificial intelligence alert is not an intervention.
A potential kidney match is not a transplant.
Each is a clinical signal. Its value depends on whether someone recognizes it, accepts responsibility, explains it to the patient, and confirms that the required action occurred.
After nine years on hemodialysis and receiving a kidney transplant in July 2025, I have experienced how easily information can become separated across doctors, hospitals, dialysis units, laboratories, pharmacies, and patients.
Recent kidney research suggests this remains a system-level problem.
When detection does not become action
A Stockholm study covering approximately 1.9 million adults annually found that acute kidney injury increased by 26 percent between 2017 and 2021.
Approximately 90 percent of identified cases were detected through creatinine changes without a recorded clinical diagnosis.
This does not prove that every case was ignored. It does expose a gap between what laboratory data revealed and what became formally recognized within the care pathway.
- Who reviewed the result?
- Was the patient informed?
- Was medication reassessed?
- Was repeat testing arranged?
- Who confirmed that follow-up occurred?
A French study involving more than 26,000 people with stage 3 to 5 chronic kidney disease found that healthcare use increased with disease severity. Older patients with stage 5 disease had greater use of general practice and non-nephrology hospital care while their contact with nephrologists declined.
Advanced kidney disease increasingly involves several providers. Without clear ownership, the patient is left trying to connect the information.
Artificial intelligence does not solve ownership
The U-Care Renal Platform uses artificial intelligence to predict moderate-to-severe acute kidney injury before onset.
Its economic model projects lower costs and improved outcomes when early warnings lead to preventive treatment. These remain projected outcomes rather than proven real-world benefits.
The larger issue is operational.
- Who receives the alert?
- How quickly must it be reviewed?
- Who can act on it?
- What happens during a shift change?
- How is the patient or family informed?
An AI system can make a correct prediction and still fail if the surrounding workflow is unreliable. Implementation should be measured by completed clinical action, rather than whether an alert appeared on a screen.
Better information flow can improve access
A kidney-allocation platform at King Fahad Specialist Hospital connected hospital and antibody data to screen the complete active waitlist during donor evaluations.
In 67 historical match runs, the automated system identified 13 potentially compatible opportunities for highly sensitized patients. The manual process identified four. Processing time fell from approximately 152 minutes to 17 minutes.
These were potential matches, rather than completed transplants. However, the study demonstrates what can happen when information flow is redesigned around a clinical objective.
The five requirements
Every kidney-care signal should have five controls.
- A named owner responsible for reviewing and advancing it.
- A defined timeframe for action.
- A patient-understandable explanation of what changed and what happens next.
- A confirmation loop showing that testing, referral, treatment, or follow-up occurred.
- An escalation process when the expected action fails.
These controls should apply whether the signal comes from a laboratory result, physician, predictive algorithm, or transplant-allocation platform.
Patient communication belongs inside the system
Patient communication is often classified as education or engagement. That understates its clinical importance.
Communication affects whether patients complete testing, understand medication changes, recognize deterioration, attend follow-up, and know whom to contact when something goes wrong.
A patient who leaves an appointment unable to explain what changed or what happens next represents a system reliability problem.
Kidney care is producing more data and stronger predictive tools. The next challenge is ensuring that every important signal becomes an understood, assigned, completed, and documented action.
Without that infrastructure, early detection remains an unfinished promise.