Page 73 - 22-0424

The International Journal of the Royal Society of Thailand
Volume XI - 2019

of false notifications may cause notification fatigue to physicians in real practice.
The prediction only valid in the sub population because they used database from
US Department of Veterans Affairs that most of the population were male (Kellum

and Bihorac, 2019; Van Biesen et al., 2019). Because AKI stage 1 events were less
clearly associated with clinical outcomes, we suggested that AI prediction in the
future research may focus on stage 2,3 AKI instead of all stage of AKI to decrease
false alert rate. Definition of AKI in this paper was defined by KDIGO 2012 criteria
by rising of creatinine level. However, creatinine is not a good marker of decreased
in GFR and renal damage (Teo and Endre, 2017). Incorporate renal biomarker to
supplement diagnosis in patients who AI labeled as at risk is a good idea and may
help to identify the early AKI patients with more precision (Zhang, 2019). In the
future, the advancement of AI may able us to predict AKI with more precision.

Improving AKI care process

Standard treatment of AKI patients follows KDIGO care bundle. The
bundle include maintaining hemodynamic status, refraining from nephron toxic
drug and avoiding of unnecessary contrast media (Khwaja, 2012). Even with
all efforts of treatment, some patients eventually progress and require renal
replacement therapy (RRT). Timing of RRT in AKI still inconclusive and several
landmark studies demonstrated heterogeneous results (Barbar et al., 2014; Gaudry
et al., 2016; Zarbock et al., 2016). Too early RRT may causes unnecessary RRT in
some patients. Moreover, the patients may develop RRT complications including
catheter-related bloodstream infection, bleeding and hypotension. On the other

hand, late RRT patients are at risk of electrolyte imbalance and fluid overload
causing increased mortality rate. In our opinion, AI can be useful in assisting
nephrologists to select appropriate timing of RRT.

Frequently used mode of RRT in critically ill patients are sustained low
efficiency hemodialysis (SLED) and continuous renal replacement therapy (CRRT)
(Bagshaw et al., 2008; Srisawat et al., 2018). Principle of SLED is similar to
hemodialysis, but used lower blood flow rate and increased therapy duration
(Berbece and Richardson, 2006; Fieghen et al., 2010; Srisawat et al., 2018). The
appropriate prescriptions of SLED including ultrafiltration setting, fluid removal
rate and electrolytes in dialysate are required to minimize complications. These can
be challenging, even for experienced nephrologists, due to complexity of critical
care patients. There are several studies aim to reduce hemodialysis complication

Phatadon Sirivongrangson
Nattachai Srisawat 67

11/7/2565 BE 13:28
_22-0424(055-076)7.indd 67 11/7/2565 BE 13:28
_22-0424(055-076)7.indd 67