Dados do Resumo

Título

ONCOASSIST: A smartphone-accessible artificial intelligence tool for predicting transient hypocalcemia after total thyroidectomy

Descrição da inciativa

Acute postoperative hypocalcemia is the most frequent complication after total thyroidectomy, occurring in up to 30% of patients. This condition results from surgical manipulation and the eventual removal of the parathyroid gland. Acute hypocalcemia can cause symptoms ranging from mild, such as tingling and muscle cramps, to severe, such as seizures and cardiac arrhythmias, representing a significant challenge in post-operative management. Inadequate management of hypocalcemia can worsen the patient experience, prolong the length of hospital stay or increase the rate of return to the emergency room.
Traditionally, the diagnosis of acute postoperative hypocalcemia has been based on clinical complaints and postoperative laboratory tests. However, clinical complaints are not always specific and laboratory tests can be altered due to the processing method and other factors such as hypoalbuminemia. These limitations can make it difficult to make an accurate diagnosis, which can lead to the improper use of calcium replacement or failure to indicate replacement when it is needed. Therefore, decision support tools that take risk into account, based on the characteristics of the patient, the extent of the surgery and the post-operative tests, can be useful in personalizing the patient's care strategy.
The application of artificial intelligence (AI) and machine learning (ML) techniques offers new opportunities to solve these limitations in clinical diagnosis, as they can quickly analyze large volumes of clinical data to identify patterns and risk factors associated with specific complications, such as acute hypocalcemia. Recent studies have shown that machine learning models outperform traditional methods in predicting clinical outcomes due to their ability to consider multiple variables and their non-linear interactions.
The integration of AI-based predictive models can provide a more accurate and robust clinical decision support system (CDSS), assisting surgeons in identifying patients at higher risk of developing hypocalcemia after total thyroidectomy. This facilitates the implementation of preventive measures and the planning of personalized interventions, improving postoperative management and patient safety.

Impacto e resultados

The development of the artificial intelligence algorithm used a database from prospective cohort study with patients undergoing total thyroidectomy with or without neck dissection. 7,066 patients who underwent total thyroidectomy, with 5462 (77.3%) females, 5,248 (74.3%) of whom had malignant pathologies and 1,003 (14.2%) of whom underwent neck dissection. The rate of transient hypocalcemia was 15.6%.This algorithm has been made available on the smartphone with the possibility of calculating the individual risk of hypocalcemia based on the patient's information compared to the historical series, making post-operative care for these patients safer. Enabling more homogeneous care and reducing the need for other support tests.