Machine learning can identify patients at risk of hyperparathyroidism without known calcium and intact parathyroid hormone

Using machine learning models for predicting monthly iPTH levels in hemodialysis patients

BACKGROUND AND OBJECTIVE

Intact parathyroid hormone (iPTH), also known as active parathyroid hormone, is an important indicator commonly for monitoring secondary hyperparathyroidism (SHPT) in patients undergoing hemodialysis. The aim of this study was to use machine learning (ML) models to predict monthly iPTH levels in patients undergoing hemodialysis.

METHODS

We conducted a retrospective study on patients undergoing regular hemodialysis. Patients' blood examinations data was collected from Taiwan Society of Nephrology - Kidney Dialysis, Transplantation (TSN-KiDiT) registration system, and patients' medications data was collected from Pingtung Christian Hospital (PTCH), Taiwan. We used five different ML models to classify patients into three distinct categories based on their iPTH levels: iPTH < 150, iPTH ≥ 150 & iPTH < 600, and iPTH ≥ 600(pg/ml).

RESULTS

We ultimately included 1,351 patients in our study and processed the data in four different ways. These methods varied based on the duration of the data (either using data from just one month or continuously over three months) and the number of features used (either all 52 features or only 20 most important features identified by SHapley Additive exPlanations (SHAP) analysis). The XGBoost model, using data from a continuous three-month period and all available features, yielded the best Weighted AUROC (0.922).

CONCLUSIONS

ML is highly effective in predicting iPTH levels in hemodialysis patients, notably accurate for those with iPTH over 600 pg/ml. This method enables early identification of high-risk patients, reducing reliance on retrospective blood test assessments. Future research should focus on advancing explainable AI methods to foster clinicians' trust, and developing adaptable ML frameworks that could seamlessly integrate with existing healthcare systems.

Potential Impact of Opportunistic CT for Closing Diagnosis Gaps in Primary Hyperparathyroidism

OBJECTIVE

Primary hyperparathyroidism (PHPT) is underdiagnosed. Opportunistic assessment for enlarged parathyroid glands on routine CT examinations is a proposed approach to improve diagnosis. However, the proportion of at-risk patients with a relevant CT is unknown. We aimed to determine the proportion of individuals with hypercalcemia untested for PHPT who had CT examinations on which opportunistic screening could have been performed and to assess characteristics associated with imaging availability.

METHODS

This retrospective study included adults with hypercalcemia untested for PHPT within our health system between January 2018 and December 2022. Each patient was classified as imaging available versus unavailable based on the presence of contrast-enhanced CTs including the parathyroid region performed between January 2013 and December 2022. Characteristics of these groups were compared.

RESULTS

The sample comprised 10,702 patients (mean age, 57 years; 6,422 female and 4,280 male patients) with CTs available in 1,318 (12.3%). Characteristics associated with the greatest odds of available CT were Charlson Comorbidity Index ≥ 5 (odds ratio [OR] 5.29, P < .0001), death during the study period (OR 2.31, P < .0001), fatigue (OR 1.90, P < .0001), weakness (OR 1.60, P < .0001), and calcium > 12.0 mg/dL (OR 1.44, P < .0001). Characteristics associated with the lowest odds of available CT were age ≥ 85 years (OR 0.27, P < .0001), age < 35 years (OR 0.58, P < .0001), and chronic kidney disease (OR 0.64, P < .0001).

CONCLUSION

More than 12% of patients with hypercalcemia who were untested for PHPT had at least one CT that could have been used to opportunistically assess the parathyroid glands. Patients with imaging tended to have more comorbidities than those without.

Testing for Primary Hyperparathyroidism in 17,491 Patients With Hypercalcemia

BACKGROUND

Primary hyperparathyroidism (PHPT) is underdiagnosed and associated with many adverse health effects. Historically, many hypercalcemic patients have not received parathyroid hormone (PTH) testing; however, underlying reasons are uncertain. Our goals are to determine the PTH testing rate among hypercalcemic individuals at a large academic health system and to assess for characteristics associated with testing versus not testing for PHPT to inform future strategies for closing testing gaps.

METHODS

This retrospective study included adult patients with ≥1 elevated serum calcium result between 2018 and 2022. Based on the presence or absence of a serum PTH result, individuals were classified as "screened" versus "unscreened" for PHPT. Demographic and clinical characteristics of these groups were compared.

RESULTS

The sample comprised 17,491 patients: 6567 male (37.5%), 10,924 female (62.5%), mean age 59 y. PTH testing was performed in 6096 (34.9%). Characteristics independently associated with the greatest odds of screening were 5+ elevated calcium results (odds ratio [OR] 5.02, P < 0.0001), chronic kidney disease (OR 3.63, P < 0.0001), maximum calcium >12.0 mg/dL (OR 2.48, P < 0.0001), and osteoporosis (OR 2.42, P < 0.0001). Characteristics associated with lowest odds of screening were age <35 y (OR 0.60, P < 0.0001), death during the study period (OR 0.68, P < 0.0001), age ≥85 y (OR 0.70, P = 0.0007), and depression (OR 0.84; P = 0.0081).

CONCLUSIONS

Only 35% of hypercalcemic patients received PTH testing. Although the presence of PHPT-associated morbidity was generally associated with increased rates of screening, hypercalcemic patients with depression were 16% less likely to be tested.

Secondary Hyperparathyroidism

Secondary hyperparathyroidism (SHPT) does not initiate as a primary dysfunction of parathyroid glands resulting from an intrinsic defect or disease but is the physiologic response of parathyroids to metabolic changes elsewhere in the body occurring over time. SHPT is a manifestation of a chronic condition that classically occurs from chronic kidney disease. In fact, given the relatively recent transition of populations from outside (agrarian) to indoor (industrial, information technology, and so forth) employment and a consequent reduction in sun exposure, combined with diets of highly processed food, vitamin D and calcium deficiencies are now the leading causes of SHPT.

Machine Learning-Based Prediction of Elevated PTH Levels Among the US General Population

CONTEXT

Although elevated parathyroid hormone (PTH) levels are associated with higher mortality risks, the evidence is limited as to when PTH is expected to be elevated and thus should be measured among the general population.

OBJECTIVE

This work aimed to build a machine learning-based prediction model of elevated PTH levels based on demographic, lifestyle, and biochemical data among US adults.

METHODS

This population-based study included adults aged 20 years or older with a measurement of serum intact PTH from the National Health and Nutrition Examination Survey (NHANES) 2003 to 2006. We used the NHANES 2003 to 2004 cohort (n = 4096) to train 6 machine-learning prediction models (logistic regression with and without splines, lasso regression, random forest, gradient-boosting machines [GBMs], and SuperLearner). Then, we used the NHANES 2005 to 2006 cohort (n = 4112) to evaluate the model performance including area under the receiver operating characteristic curve (AUC).

RESULTS

Of 8208 US adults, 753 (9.2%) showed PTH greater than 74 pg/mL. Across 6 algorithms, the highest AUC was observed among random forest (AUC [95% CI] = 0.79 [0.76-0.81]), GBM (AUC [95% CI] = 0.78 [0.75-0.81]), and SuperLearner (AUC [95% CI] = 0.79 [0.76-0.81]). The AUC improved from 0.69 to 0.77 when we added cubic splines for the estimated glomerular filtration rate (eGFR) in the logistic regression models. Logistic regression models with splines showed the best calibration performance (calibration slope [95% CI] = 0.96 [0.86-1.06]), while other algorithms were less calibrated. Among all covariates included, eGFR was the most important predictor of the random forest model and GBM.

CONCLUSION

In this nationally representative data in the United States, we developed a prediction model that potentially helps us to make accurate and early detection of elevated PTH in general clinical practice. Future studies are warranted to assess whether this prediction tool for elevated PTH would improve adverse health outcomes.