SPONTANEOUS STONE PASSAGE
Ureteral stones 4 mm or less , particularly in the distal ureter , with no urinary tract infection and with minimal pain , can safely be observed with 95 % chance of stone passage . Solakhan and coworkers 8 built an artificial neural network ( ANN ) to predict stone passage in the range of 97 %. The authors pointed out that the stone pass rate is mainly dependent on stone size , which is a linear predictor ; the results are the same using either imaging or ANN models .
EXTRACORPOREAL SHOCKWAVE LITHOTRIPSY ( ESWL )
ESWL is a popular noninvasive approach for urinary stones . It is used for stones of various sizes and locations with variable success rates with minimal complications . Recently , patient and stone parameters have been shown to predict stone free outcomes . These include but are not limited to stone size , location , composition and skin to stone distance . AI can be applied to ESWL to select patients and forecast outcomes . Michaels et al 9 studied 98 patients who underwent ESWL for kidney and ureteral stones . The ANN was developed with 16 input variables and this model yielded 91 % success rate . This has been supported by Seckiner and coworkers 10 ANN showing 99.3 % prediction accuracy .
PERCUTANEOUS NEPHROLITHOTOMY ( PCNL )
PCNL is favored in large complex stones and staghorn calculi . Of all the modalities , it is very invasive and prone to complications . Despite its invasiveness , it is very successful in eliminating the stone burden . Validated nomograms such as the Clinical Research Office of the Endourology Society score ( CROES ) are popular in predicting success rates . The CROES nomogram is based on 2,806 patients and a logistic regression model showed a stone free status (< 4 mm fragments ) is 64.8 %. Aminsharifi et al 11 developed an ANN with pre & post-operative data ( 16 variables in 200 patients ), and a test group of 254 patients who had PCNL . Their algorithm predicted 83 % stone free rate , 86 % blood transfusion rate . Other centers reported similar results using machine-learning models in the range of 70-94.8 %.
STONE PREVENTION
Most practicing urologists focus on treatment of stone episodes and very few are knowledgeable or interested in stone prevention . Recurrent stone formers experience poor quality of life and risk addiction to opioids . Management of nephrolithiasis needs a multidisciplinary team of urologists , nephrologists , dietitians and counselors . Metabolic stone workup is indicated in patients with family history , bilateral stones , multiple stones , stones as children and frequent stone episodes over a short time . This includes stone analysis , 24-hour urine analysis for risk factors and dietary review . Combined data sets could be fed into algorithms and stone recurrence rate could be individualized to patients .
UROLOGY : CLINICAL UPDATES FROM THE PRACTICE FUTURE DIRECTIONS
In addition to stone prevention strategies , AI could be used in directing the PCNL track formation to help urologists find the best route possible for optimum results . Many studies are developed based on “ big data ” and most of them are very new and not externally validated . However , we predict that these models will be applied more often in the future .
CONCLUSION
AI could be used in the diagnosis , management and prevention of urinary stone disease to improve quality , safety and patient satisfaction . As more data sets are available through electronic medical records , AI will be used more in the everyday practice of clinical medicine , ideally customized for the individual patient .
MAIN POINTS
1 . AI can be used more in diagnosis of the urinary stone disease .
2 . AI can help identify stone composition to design treatment approach .
3 . AI can help predict stone free rates after PCNL , ESWL and stone passage rates in expectant management . 4 . AI can aid stone prevention strategies .
References
1
Frankish K , Ramsey WM . The Cambridge handbook of artificial intelligence . Cambridge , UK : Cambridge University Press ; 2014 .
2
Fulgham PF , Assimos DG , Pearle MS , Preminger GM . Clinical effectiveness protocols for imaging in the management of ureteral calculous disease : AUA technology assessment . J Urol 2013 ; 189:1203 – 1213 .
3
Längkvist M , Jendeberg J , Thunberg P , et al . Computer aided detection of ureteral stones in thin slice computed tomography volumes using Convolutional Neural Networks . Comput Biol Med 2018 ; 97:153 – 160 .
4
Parakh A , Lee H , Lee JH , et al . Urinary stone detection on CT images using deep convolutional neural networks : evaluation of model performance and generalization . Radiol Artif Intell 2019 ; 1 : e180066 .
5
Serrat J , Lumbreras F , Blanco F , et al . myStone : A system for automatic kidney stone classification . Expert Syst Appl 2017 ; 89:41 – 51 .
6
Black KM , Law H , Aldoukhi A , et al . Deep learning computer vision algorithm for detecting kidney stone composition . BJU Int 2020 ; 125:920 – 924 .
7
Kazemi Y , Mirroshandel SA . A novel method for predicting kidney stone type using ensemble learning . Artif Intell Med 2018 ; 84:117 – 126 .
8
Solakhan M , Seckiner SU , Seckiner I . A neural network-based algorithm for predicting the spontaneous passage of ureteral stones . Urolithiasis 2020 ; 48:527 – 532
9
Michaels EK , Niederberger CS , Golden RM , et al . Use of a neural network to predict stone growth after shock wave lithotripsy . Urology 1998 ; 51:335 – 338 .
10
Seckiner I , Seckiner S , Sen H , et al . A neural network – based algorithm for predicting stone - free status after ESWL therapy . Int Braz J Urol 2017 ; 43:1110 – 1114 .
11
Aminsharifi A , Irani D , Pooyesh S , et al . Artificial neural network system to predict the postoperative outcome of percutaneous nephrolithotomy . J Endourol 2017 ; 31:461 – 467 .
Dr . Rai is a resident in the urology residency program at the University of Louisville .
Dr . Ankem is a Professor & Chair of the Department of Urology at the University of Louisville School of Medicine .
AUGUST 2021 11