Researchers has applied artificial intelligence to the problem of detecting acute kidney injury in hospitalized patients


A team of researchers from DeepMind, the U.S. Veterans Administration and several other institutions in the U.K. and the U.S. has applied artificial intelligence to the problem of detecting acute kidney injury in hospitalized patients.

In their paper published in the journal Nature, the group describes their deep learning project and how well it performed.

Acute kidney injury (AKI) is where something happens in the body that leads to deterioration of the kidneys.

In severe cases, it can lead to the need for a transplant or death.

AKI typically occurs with patients being cared for in a hospital and is usually a sign of a rapid downturn that requires emergency measures by hospital staff to prevent further irreversible kidney damage.

In this new effort, the researchers wondered if it might be possible to use AI to detect signs of AKI earlier than normally occurs in a hospital, thus giving patients a much better outcome.

To find out, the researchers worked with the VA, which runs multiple hospitals for veteran care across the United States.

The effort involved feeding a deep learning system the health records of 703,782 veterans ranging in age from 18 to 90 – all of whom had suffered from some form of AKI.

The system used the data from the veterans to detect patterns of small changes in the patients, such as creatinine levels in the blood.

The team then reran the data as a way to test their system, to see how well it could predict AKI in the same patients.

The researchers report that their system worked remarkably well for patients who developed the most serious forms of AKI – it correctly predicted them for approximately 90 percent of cases (with a lead time of 48 hours).

It did less well for less serious cases – for all of the cases tested, the system was able to correctly predict an AKI event in just 55.8 percent of cases.

It also gave two false positives for every correct result. Still, the researchers are optimistic about the possibility of using AI in many types of critical care scenarios such as the likelihood of heart attack.

The researchers are planning to continue their research – they hope to expand the study to a broader population.

The concept of Acute Renal Failure (ARF)1 has undergone significant re-examination in recent years.

Traditionally, emphasis was given to the most severe acute reduction in kidney function, as manifested by severe azotaemia and often by oliguria or anuria.

However, recent evidence suggests that even relatively mild injury or impairment of kidney function manifested by small changes in serum creatinine (sCr) and/or urine output (UO), is a predictor of serious clinical consequences.25

Acute Kidney Injury (AKI) is the term that has recently replaced the term ARF. AKI is defined as an abrupt (within hours) decrease in kidney function, which encompasses both injury (structural damage) and impairment (loss of function).

It is a syndrome that rarely has a sole and distinct pathophysiology. Many patients with AKI have a mixed aetiology where the presence of sepsis, ischaemia and nephrotoxicity often co-exist and complicate recognition and treatment.

Furthermore the syndrome is quite common among patients without critical illness and it is essential that health care professionals, particularly those without specialisation in renal disorders, detect it easily.

Classification of AKI includes pre-renal AKI, acute post-renal obstructive nephropathy and intrinsic acute kidney diseases.

Of these, only ‘intrinsic’ AKI represents true kidney disease, while pre-renal and post-renal AKI are the consequence of extra-renal diseases leading to the decreased glomerular filtration rate (GFR).

If these pre- and/or post-renal conditions persist, they will eventually evolve to renal cellular damage and hence intrinsic renal disease.

The current diagnostic approach of AKI is based on an acute decrease of GFR, as reflected by an acute rise in sCr levels and/or a decline in UO over a given time interval.68 

Recently several biomarkers have been proposed for the diagnosis of AKI and these are in various stages of development and validation.912 

Nevertheless, it is not clear, if a single or multiple biomarker approach is necessary to diagnose the complicated and multifactorial aspects of AKI.1316

However, in addition to the analytical difficulties associated with each specific biomarker, there is also an issue concerning the appropriate reference point, and more specifically about using sCr as the standard, for the clinical evaluation of these biomarkers.

It is known that sCr is insensitive to acute changes of renal function and levels can vary widely with age, gender, muscle mass, diet, medications and hydration status. Moreover it is not a direct marker of tubular damage, but rather a marker of GFR, and substantial increases in sCr can be observed in renal hypo-perfusion even when the kidneys are structurally intact, resulting in pre-renal azotaemia.

For these reasons sCr is considered an ‘imperfect gold standard’ for the diagnosis of AKI.17 Another issue with sCr is that in most clinical situations its true baseline value is not known, which makes the evaluation of patients very difficult.1820 

Moreover, given the phenotypic variability of AKI (different clinical phenotypes with distinct underlying pathophysiologies), it is not clear whether different approaches are necessary for diagnosis and monitoring of the clinical course and therapy.

In this review we will discuss the epidemiology and the definition of AKI. We will also discuss the clinical phenotypes, their pathophysiology and the link between AKI and remote organ dysfunction.

More information: Nenad Tomašev et al. A clinically applicable approach to continuous prediction of future acute kidney injury, Nature (2019). DOI: 10.1038/s41586-019-1390-1

Journal information: Nature


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.