Chronic Kidney Disease (CKD) is rising at an alarming rate worldwide. The kidney's primary function is to maintain fluid and electrolyte balance. Any changes in renal function, whether acute or chronic, can cause multiple imbalances. In many cases, electrolyte imbalance, particularly potassium imbalance, has resulted in sudden cardiac deaths in such patients. Currently blood tests are conducted for measuring the electrolytes in patients. However continuous monitoring of imbalance or ease of such a test at home is not possible leading to life threatening conditions. Recent studies have found that electrolytes imbalance can be detected using ECG signals. ECG are commonly acquired during clinical examination and can now be easily acquired by many wearable sensors used for fitness and monitoring. ECG Interpretation requires expertise, however interpretation becomes difficult in cases where large amount of ECG data is being continuously generated by wearable sensors. Automatic interpretation of such ECG data would be useful especially in patients suffering from cardiac abnormalities. Machine Learning is a branch of Artificial Intelligence that allows computers to make accurate predictions. When compared to traditional or manual methods, the use of machine learning techniques, particularly deep learning, in ECG interpretation has demonstrated encouraging outcomes. In this review, we discuss the problem of electrolyte imbalance, explore the potential of ECG as a diagnostic tool and present the recent developments in using machine learning techniques especially deep learning for electrolyte imbalance detection using ECG. Further this paper also discusses the problem of interpretability of deep learning models and potential solutions offered by a relatively new field called Explainable AI. Finally the paper discusses the challenges faced by researchers in electrolyte imbalance detection using ECG

Chronic Kidney Disease; Electrolyte Imbalance; Machine Learning; Deep Learning; ECG