Predicting Kidney Discard Using Machine Learning

Patient and Caregiver Perceptions on the Allocation Process and Waitlist, and Accepting a Less-Than-Ideal Kidney: A Canadian Survey

Background

Transplanting less-than-ideal (LTI) kidneys could help optimize organ utilization, but little is known about how patients and caregivers perceive the allocation process, waitlist, or LTI kidneys.

Objective

To explore the perspectives of patients and caregivers on the Canadian kidney transplant allocation process, waitlist, and LTI kidneys.

Design

Electronic survey.

Setting

Canada.

Patients

Transplant recipients, candidates, and caregivers.

Methods

A bilingual electronic national survey was administered from January to March 2024. The questionnaire contained sections on demographics, perceptions of organ allocation and acceptance, LTI kidneys, and educational preferences. Descriptive analysis was performed.

Results

Two hundred fifty-one responses were analyzed, including patients (63%, n = 159), and caregivers (37%, n = 92), from 11 provinces and territories. Three-quarters (74%, n = 186) understood how patients are placed on the waiting list, and 65% (n = 162) understood how donor kidneys are allocated, but 72% (n = 181) and 68% (n = 171) wanted more information about the waitlist and donor kidney allocation criteria, respectively. Approximately 20% felt that the waitlist and allocation processes were not transparent. Awareness about the option to refuse a deceased donor kidney offer was high (69%, n = 174), yet nearly half of respondents (46%, n = 115) expressed concern about being disadvantaged if an offer for a deceased donor kidney was refused. One-third of participants (33%, n = 83) were open to accepting an LTI kidney.

Limitations

Compared to the general population, more study participants were white, and the majority were educated and financially at ease. This limits the generalizability of the results.

Conclusion

Enhanced communication is required to improve transparency and information about the allocation system and waitlist in Canada.

Advancements in Artificial Intelligence for Kidney Transplantology: A Comprehensive Review of Current Applications and Predictive Models

Background: Artificial intelligence is rapidly advancing within the domains of medicine and transplantology. In this comprehensive review, we provide an in-depth exploration of current AI methodologies, with a particular emphasis on machine learning and deep learning techniques, and their diverse subtypes. These technologies are revolutionizing how data are processed, analyzed, and applied in clinical decision making. Methods: A meticulous literature review was conducted with a focus on the application of artificial intelligence in kidney transplantation. Four research questions were formulated to establish the aim of the review. Results: We thoroughly examined the general applications of AI in the medical field, such as feature selection, dimensionality reduction, and clustering, which serve as foundational tools for complex data analysis. This includes the development of predictive models for transplant rejection, the optimization of personalized immunosuppressive therapies, the algorithmic matching of donors and recipients based on multidimensional criteria, and the sophisticated analysis of histopathological images to improve the diagnostic accuracy. Moreover, we present a detailed comparison of existing AI-based algorithms designed to predict kidney graft survival in transplant recipients. In this context, we focus on the variables incorporated into these predictive models, providing a critical analysis of their relative importance and contribution to model performance. Conclusions: This review highlights the significant advancements made possible through AI and underscores its potential to enhance both clinical outcomes and the precision of medical interventions in the field of transplantology.

Predictive value of machine learning model based on CT values for urinary tract infection stones

Preoperative diagnosis of infection stones presents a significant clinical challenge. We developed a machine learning model to predict urinary infection stones using computed tomography (CT) values, enabling in vivo preoperative identification. In this study, we included 1209 patients who underwent urinary lithotripsy at our hospital. Seven machine learning algorithms along with eleven preoperative variables were used to construct the prediction model. Subsequently, model performance was evaluated by calculating AUC and AUPR for subjects in the validation set. On the validation set, all seven machine learning models demonstrated strong discrimination (AUC: 0.687-0.947). Additionally, the XGBoost model was identified as the optimal model significantly outperforming the traditional LR model. Taken together, the XGBoost model is the first machine learning model for preoperative prediction of infection stones based on CT values. It can rapidly and accurately identify infection stones in vitro, providing valuable guidance for urologists in managing these stones.

Present and Future Applications of Artificial Intelligence in Kidney Transplantation

Artificial intelligence (AI) has a wide and increasing range of applications across various sectors. In medicine, AI has already made an impact in numerous fields, rapidly transforming healthcare delivery through its growing applications in diagnosis, treatment and overall patient care. Equally, AI is swiftly and essentially transforming the landscape of kidney transplantation (KT), offering innovative solutions for longstanding problems that have eluded resolution through traditional approaches outside its spectrum. The purpose of this review is to explore the present and future applications of artificial intelligence in KT, with a focus on pre-transplant evaluation, surgical assistance, outcomes and post-transplant care. We discuss its great potential and the inevitable limitations that accompany these technologies. We conclude that by fostering collaboration between AI technologies and medical practitioners, we can pave the way for a future where advanced, personalised care becomes the standard in KT and beyond.

Impact of donor smoking history on kidney transplant recipient outcomes: A systematic review and meta-analysis

BACKGROUND

Impact of donor smoking history on kidney transplant recipient outcomes is undefined.

METHODS

We systematically searched, critically appraised, and summarized associations between donor smoking and primary outcomes of death-censored and all-cause graft failure (DCGF, ACGF), and secondary outcomes of allograft histology, delayed graft function, serum creatinine, estimated glomerular filtration rate, and mortality. We searched MEDLINE, Embase, and Cochrane Databases from 2000 to 2023. Risk of bias was assessed using Risk of Bias in Non-randomized Studies - of Exposure tool. Quality of evidence was assessed by Grading of Recommendations Assessment, Development and Evaluation Working Group recommendations. We pooled results using inverse variance, random-effects model and reported hazard ratios for time-to-event outcomes or binomial proportions. Statistical heterogeneity was assessed with I2 statistic.

RESULTS

From 1785 citations, we included 17 studies. Donor smoking was associated with modestly increased DCGF (HR 1.05 (95% CI: 1.01, 1.09); I2 = 0%; low quality of evidence), predominantly in deceased donors, and ACGF in adjusted analyses (HR 1.12 (95% CI: 1.06, 1.19); I2 = 20%; very low quality of evidence). Other outcomes could not be pooled meaningfully.

CONCLUSIONS

Kidney donor smoking history was associated with modestly increased risk of death-censored graft failure and all-cause graft failure. This review emphasizes the need for further research, standardized reporting, and thoughtful consideration of donor factors like smoking in clinical decision-making on kidney utilization and allocation.

Preoperative Risk Assessment of Early Kidney Graft Loss

Background

A large proportion of potential organ donors are not utilized for kidney transplantation out of risk of early allograft loss because of donor-related characteristics. These can be summarized using kidney donor profile index (KDPI). Because KDPI affects the choice of the recipient, the predictive ability of KDPI is tied to recipient attributes. These have been questioned to explain most of the predictive ability of KDPI. This study aims to quantify the effect of the donor on early graft loss (EGL) by accounting for nonrandom allocation.

Methods

This study included patients undergoing kidney transplantation from deceased donors between 2014 and 2020 from the Scientific Registry of Transplantation Recipients. EGL, defined as a return to dialysis or retransplantation during the first posttransplant year, was the primary endpoint. Nonrandom allocation and donor-recipient matching by KDPI necessitated the use of inverse probability treatment weighting, which served to assess the effect of KDPI and mitigate selection bias in a weighted Cox regression model.

Results

The study comprised 89 290 transplantations in 88 720 individual patients. Inverse probability treatment weighting resulted in a good balance of recipient covariates across values of continuous KDPI. Weighted analysis showed KDPI to be a significant predictor for short-term outcomes. A comparable (in terms of age, time on dialysis, previous transplants, gender, diabetes status, computed panel-reactive antibodies, and HLA mismatches) average recipient, receiving a kidney from a donor with KDPI 40-60 had a 3.5% risk of EGL increased to a risk of 7.5% if received a kidney from a KDPI >95 donor (hazard ratio, 2.3; 95% confidence interval, 1.9-2.7). However, for all-cause survival KDPI was less influential.

Conclusions

The predictive ability of KDPI does not stem from recipient confounding alone. In this large sample-sized study, modeling methods accounting for nonindependence of recipient selection verify graft quality to effectively predict short-term transplantation outcomes.

Development and validation of a diagnostic model to differentiate spinal tuberculosis from pyogenic spondylitis by combining multiple machine learning algorithms

This study focused on the development and validation of a diagnostic model to differentiate between spinal tuberculosis (STB) and pyogenic spondylitis (PS). We analyzed a total of 387 confirmed cases, out of which 241 were diagnosed with STB and 146 were diagnosed with PS. These cases were randomly divided into a training group (n = 271) and a validation group (n = 116). Within the training group, four machine learning (ML) algorithms (least absolute shrinkage and selection operator [LASSO], logistic regression analysis, random forest, and support vector machine recursive feature elimination [SVM-RFE]) were employed to identify distinctive variables. These specific variables were then utilized to construct a diagnostic model. The model's performance was subsequently assessed using the receiver operating characteristic (ROC) curves and the calibration curves. Finally, internal validation of the model was undertaken in the validation group. Our findings indicate that PS patients had an average platelet-to-neutrophil ratio (PNR) of 277.86, which was significantly higher than the STB patients' average of 69.88. The average age of PS patients was 54.71 years, older than the 48 years recorded for STB patients. Notably, the neutrophil-to-lymphocyte ratio (NLR) was higher in PS patients at 6.15, compared to the 3.46 NLR in STB patients. Additionally, the platelet volume distribution width (PDW) in PS patients was 0.2, compared to 0.15 in STB patients. Conversely, the mean platelet volume (MPV) was lower in PS patients at an average of 4.41, whereas STB patients averaged 8.31. Hemoglobin (HGB) levels were lower in PS patients at an average of 113.31 compared to STB patients' average of 121.64. Furthermore, the average red blood cell (RBC) count was 4.26 in PS patients, which was less than the 4.58 average observed in STB patients. After evaluation, seven key factors were identified using the four ML algorithms, forming the basis of our diagnostic model. The training and validation groups yielded area under the curve (AUC) values of 0.841 and 0.83, respectively. The calibration curves demonstrated a high alignment between the nomogram-predicted values and the actual measurements. The decision curve indicated optimal model performance with a threshold set between 2% and 88%. In conclusion, our model offers healthcare practitioners a reliable tool to efficiently and precisely differentiate between STB and PS, thereby facilitating swift and accurate diagnoses.

Evaluation of Opening Offers Early for Deceased Donor Kidneys at Risk of Nonutilization

BACKGROUND

Reducing nonutilization of kidneys recovered from deceased donors is a current policy concern for kidney allocation in the United States. The likelihood of nonutilization is greater with a higher kidney donor risk index (KDRI) offer. We examine how opening offers for organs with KDRI >1.75 to the broader waitlist at varying points of time affects usage rates.

METHODS

We simulate kidney allocation using data for January 2018 to June 2019 from Organ Procurement and Transplantation Network. For the simulation experiment, allocation policy is modified so that KDRI >1.75 organs are offered to all local candidates (same donation service area) after a set amount of cold time simultaneously. Open offers to candidates nationally are similarly examined.

RESULTS

Simulation results ( n =50 replications) estimate that opening offers locally for KDRI >1.75 after 10 hours yields a nonutilization rate of 38% (range: 35%-42%), less than the prevailing rate of 55% of KDRI >1.75 kidneys. Opening offers after 5 hours yields 30% (range: 26%-34%), reducing the prevailing nonutilization rate by 45%. Opening offers nationally after 10 and 5 hours yields nonutilization rates of 11% (range: 8%-15%) and 6% (range: 4%-9%) for KDRI >1.75 kidneys, respectively.

CONCLUSIONS

Simulation findings indicate that opening offers and adjusting their timing can significantly reduce nonutilization of high-KDRI kidneys.

Deceased donor kidneys from higher distressed communities are significantly less likely to be utilized for transplantation

The proportion of kidneys procured for transplantation but not utilized exceeds 20% in the United States. Factors associated with nonutilization are complex, and further understanding of novel causes are critically important. We used the national Scientific Registry of Transplant Recipients data (2010-2022) to evaluate associations of Distressed Community Index (DCI) of deceased donor residence and likelihood of kidney nonutilization (n = 209 413). Deceased donors from higher distressed communities were younger, had an increased history of hypertension and diabetes, were CDC high-risk, and had higher terminal creatinine and donation after brain death. Mechanisms and circumstances of death varied significantly by DCI. The proportion of kidney nonutilization was 19.9%, which increased by DCI quintile (Q1 = 18.1% to Q5 = 21.6%). The adjusted odds ratio of nonutilization from the highest quintile DCI communities was 1.22 (95% CI = 1.16-1.28; reference = lowest DCI), which persisted stratified by donor race. Donors from highly distressed communities were highly variable by the donor service area (range: 1%-51%; median = 21%). There was no increased risk for delayed graft function or death-censored graft loss by donor DCI but modest increased adjusted hazard for overall graft loss (high DCI = 1.05; 95% CI = 1.01-1.10; reference = lowest DCI). Results indicate that donor residential distress is associated with significantly higher rates of donor kidney nonutilization with notable regional variation and minimal impact on recipient outcomes.

Prediction model of ocular metastasis from primary liver cancer: Machine learning-based development and interpretation study

BACKGROUND

Ocular metastasis (OM) is a rare metastatic site of primary liver cancer (PLC). The purpose of this study was to establish a clinical predictive model of OM in PLC patients based on machine learning (ML).

METHODS

We retrospectively collected the clinical data of 1540 PLC patients and divided it into a training set and an internal test set in a 7:3 proportion. PLC patients were divided into OM and non-ocular metastasis (NOM) groups, and univariate logistic regression analysis was performed between the two groups. The variables with univariate logistic analysis p < 0.05 were selected for the ML model. We constructed six ML models, which were internally verified by 10-fold cross-validation. The prediction performance of each ML model was evaluated by receiver operating characteristic curves (ROCs). We also constructed a web calculator based on the optimal performance ML model to personalize the risk probability for OM.

RESULTS

Six variables were selected for the ML model. The extreme gradient boost (XGB) ML model achieved the optimal differential diagnosis ability, with an area under the curve (AUC) = 0.993, accuracy = 0.992, sensitivity = 0.998, and specificity = 0.984. Based on these results, an online web calculator was constructed by using the XGB ML model to help clinicians diagnose and treat the risk probability of OM in PLC patients. Finally, the Shapley additive explanations (SHAP) library was used to obtain the six most important risk factors for OM in PLC patients: CA125, ALP, AFP, TG, CA199, and CEA.

CONCLUSION

We used the XGB model to establish a risk prediction model of OM in PLC patients. The predictive model can help identify PLC patients with a high risk of OM, provide early and personalized diagnosis and treatment, reduce the poor prognosis of OM patients, and improve the quality of life of PLC patients.

The utility of machine learning for predicting donor discard in abdominal transplantation

BACKGROUND

Increasing access and better allocation of organs in the field of transplantation is a critical problem in clinical care. Limitations exist in accurately predicting allograft discard. Potential exists for machine learning to provide a balanced assessment of the potential for an organ to be used in a transplantation procedure.

METHODS

We accessed and utilized all available deceased donor United Network for Organ Sharing data from 1987 to 2020. With these data, we evaluated the performance of multiple machine learning methods for predicting organ use. The machine learning methods trialed included XGBoost, random forest, Naïve Bayes (NB), logistic regression, and fully connected feedforward neural network classifier methods. The top two methods, XGBoost and random forest, were fully developed using 10-fold cross-validation and Bayesian optimization of hyperparameters.

RESULTS

The top performing model at predicting liver organ use was an XGBoost model which achieved an AUC-ROC of .925, an AUC-PR of .868, and an F1 statistic of .756. The top performing model for predicting kidney organ use classification was an XGBoost model which achieved an AUC-ROC of .952, and AUC-PR of .883, and an F1 statistic of .786.

CONCLUSIONS

The XGBoost method demonstrated a significant improvement in predicting donor allograft discard for both kidney and livers in solid organ transplantation procedures. Machine learning methods are well suited to be incorporated into the clinical workflow; they can provide robust quantitative predictions and meaningful data insights for clinician consideration and transplantation decision-making.

Kidney organ allocation: reducing discards

PURPOSE OF REVIEW

The donation and kidney transplant system in the United States is challenged with reducing the number of kidneys that are procured for transplant but ultimately discarded. That number can reach 20% of donated kidneys each year.

RECENT FINDINGS

The reasons for these discards, in the face of overwhelming demand, are multiple.

SUMMARY

The authors review the data supporting a number of potential causes for high discard rates as well as provide potential solutions to the problem.

An Improved CatBoost-Based Classification Model for Ecological Suitability of Blueberries

Selecting the best planting area for blueberries is an essential issue in agriculture. To better improve the effectiveness of blueberry cultivation, a machine learning-based classification model for blueberry ecological suitability was proposed for the first time and its validation was conducted by using multi-source environmental features data in this paper. The sparrow search algorithm (SSA) was adopted to optimize the CatBoost model and classify the ecological suitability of blueberries based on the selection of data features. Firstly, the Borderline-SMOTE algorithm was used to balance the number of positive and negative samples. The Variance Inflation Factor and information gain methods were applied to filter out the factors affecting the growth of blueberries. Subsequently, the processed data were fed into the CatBoost for training, and the parameters of the CatBoost were optimized to obtain the optimal model using SSA. Finally, the SSA-CatBoost model was adopted to classify the ecological suitability of blueberries and output the suitability types. Taking a study on a blueberry plantation in Majiang County, Guizhou Province, China as an example, the findings demonstrate that the AUC value of the SSA-CatBoost-based blueberry ecological suitability model is 0.921, which is 2.68% higher than that of the CatBoost (AUC = 0.897) and is significantly higher than Logistic Regression (AUC = 0.855), Support Vector Machine (AUC = 0.864), and Random Forest (AUC = 0.875). Furthermore, the ecological suitability of blueberries in Majiang County is mapped according to the classification results of different models. When comparing the actual blueberry cultivation situation in Majiang County, the classification results of the SSA-CatBoost model proposed in this paper matches best with the real blueberry cultivation situation in Majiang County, which is of a high reference value for the selection of blueberry cultivation sites.

Predicting older-donor kidneys' post-transplant renal function using pre-transplant data

This paper provides a methodology for predicting post-transplant kidney function, that is, the 1-year post-transplant estimated Glomerular Filtration Rate (eGFR-1) for each donor-candidate pair. We apply customized machine-learning algorithms to pre-transplant donor and recipient data to determine the probability of achieving an eGFR-1 of at least 30 ml/min. This threshold was chosen because there is insufficient survival benefit if the kidney fails to generate an eGFR-1 ≥ 30 ml/min. For some donor-candidate pairs, the developed algorithm provides highly accurate predictions. For others, limitations of previous transplants' data results in noisier predictions. However, because the same kidney is offered to many candidates, we identify those pairs for whom the predictions are highly accurate. Out of 6977 discarded older-donor kidneys that were a match with at least one transplanted kidney, 5282 had one or more identified candidate, who were offered that kidney, did not accept any other offer, and would have had ≥80% chance of achieving eGFR-1 ≥ 30 ml/min, had the kidney been transplanted. We also show that transplants with ≥80% chance of achieving eGFR-1 ≥ 30 ml/min and that survive 1 year have higher 10-year death-censored graft survival probabilities than all older-donor transplants that survive 1 year (73.61% vs. 70.48%, respectively).

Kidney utility and futility

Changes in kidney allocation coupled with the COVID-19 pandemic have placed tremendous strain on current systems of organ distribution and logistics. Although the number of deceased donors continues to rise annually in the United States, the proportion of marginal deceased donors (MDDs) is disproportionately growing. Cold ischemia times and kidney discard rates are rising in part related to inadequate planning, resources, and shortages. Complexity in kidney allocation and distribution has contributed to this dilemma. Logistical issues and the ability to reperfuse the kidney within acceptable time constraints increasingly determine clinical decision-making for organ acceptance. We have a good understanding of the phenotype of "hard to place" MDD kidneys, yet continue to promote a "one size fits all" approach to organ allocation. Allocation and transportation systems need to be agile, mobile, and flexible in order to accommodate the expanding numbers of MDD organs. By identifying "hard to place" MDD kidneys early and implementing a "fast-track" or open offer policy to expedite placement, the utilization rate of MDDs would improve dramatically. Organ allocation and distribution based on location, motivation, and innovation must lead the way. In the absence of change, we are sacrificing utility for futility and discard rates will continue to escalate.

Implications of Accumulated Cold Time for US Kidney Transplantation Offer Acceptance

BACKGROUND AND OBJECTIVES

Reducing discard is important for the US transplantation system because nearly 20% of the deceased donor kidneys are discarded. One cause for the discards is the avoidance of protracted cold ischemia times. Extended cold ischemia times at transplant are associated with additional risk of graft failure and patient mortality. A preference for local (within the same donor service area) or low-Kidney Donor Risk Index organs, the endogeneity of cold ischemia time during organ allocation, and the use of provisional offers all complicate the analysis of cold ischemia times' influence on kidney acceptance decision making.

DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS

Using January 2018 to June 2019 Organ Procurement and Transplantation Network data, we modeled the probability of accepting an offer for a kidney after provisional acceptance. We use logistic regression that includes cold ischemia time, Kidney Donor Risk Index, and other covariates selected from literature. Endogeneity of cold ischemia time was treated by a two-stage instrumental variables approach.

RESULTS

Logistic regression results for 3.33 million provisional acceptances from 12,369 donors and 108,313 candidates quantify trade-offs between cold ischemia time at the time of offer acceptance and donor-recipient characteristics. Overall, each additional 2 hours of cold ischemia time affected acceptance for nonlocal and local recipients (odds ratio, 0.75; 95% confidence interval, 0.73 to 0.77, odds ratio, 0.88; 95% confidence interval, 0.86 to 0.91; P<0.001). For Kidney Donor Risk Index >1.75 (Kidney Donor Profile Index >85) kidneys, an additional 2 hours of cold ischemia time for nonlocal and local recipients was associated with acceptance with odds ratio, 0.58; 95% confidence interval, 0.54 to 0.63 (nonlocal) and odds ratio, 0.65; 95% confidence interval, 0.6 to 0.7 (local); P<0.001. The effect of an additional 2 hours of cold ischemia time on acceptance of kidneys with Kidney Donor Risk Index ≤1.75 (Kidney Donor Profile Index ≤85) was less pronounced for nonlocal offers (odds ratio, 0.82; 95% confidence interval, 0.80 to 0.85; P<0.001) and not significant for local offers.

CONCLUSIONS

The acceptability of marginal organs was higher when placements were nearer to the donor and when cold ischemia time was shorter.

Dual kidney transplants from adult marginal donors: Review and perspective

The practice of dual kidney transplantation (DKT) from adult marginal deceased donors (MDDs) dates back to the mid-1990s with initial pioneering experiences reported by the Stanford and Maryland groups, at which time the primary indication was estimated insufficient nephron mass from older donors. Multiple subsequent studies of short and long-term success have been reported focusing on three major aspects of DKT: Identifying appropriate selection criteria and developing scoring systems based on pre- and post-donation factors; refining technical aspects; and analyzing mid-term outcomes. The number of adult DKTs performed in the United States has declined in the past decade and only about 60 are performed annually. For adult deceased donor kidneys meeting double allocation criteria, >60% are ultimately not transplanted. Deceased donors with limited renal functional capacity represent a large proportion of potential kidneys doomed to either discard or non-recovery. However, DKT may reduce organ discard and optimize the use of kidneys from MDDs. In an attempt to promote utilization of MDD kidneys, the United Network for Organ Sharing introduced new allocation guidelines pursuant to DKT in 2019. The purpose of this review is to chronicle the history of DKT and identify opportunities to improve utilization of MDD kidneys through DKT. This article is protected by copyright. All rights reserved.