The aim of this study is to build predictive models based on machine learning techniques and to identify risk factors associated with the development of acute mastitis in lactating women. The dataset used in this study comes from a case-control study that includes 369 breastfeeding women diagnosed with acute mastitis and 447 breastfeeding women without this condition. Data were collected from these participants that covered a variety of potential risk indicators, including age, primitive status, history of breast surgery, nipple trauma, external breast trauma, postpartum status, history of gestational diabetes, and abnormal nipple discharge.
Machine learning technologies, especially Xgboost and MLP, are used to build predictive models. Xgboost is a powerful variant of gradient boosting, known for its excellent performance characteristics, including high accuracy, robustness, scalability, and customizability18. On the other hand, MLP is an artificial neural network suitable for capturing complex patterns through multilayered nonlinear transformations, and is particularly good at learning complex relationships with high-dimensional data.16. This feature makes MLP an efficient tool for regression and classification tasks, often outperforming traditional algorithms in terms of flexibility and predictive ability19.
The application of these advanced machine learning methods, particularly in the field of medical diagnosis in the prediction and diagnosis of acute mastitis, demonstrates the ability to process complex high-dimensional datasets and uncover underlying disease patterns.20,21. For example, XGBoost can assist clinicians by analyzing clinical symptoms, diagnostic results, and imaging data, improving the accuracy and timeliness of mastitis diagnosis.twenty two. Meanwhile, MLP can reveal nonlinear relationships in a wide range of datasets including patient history and genetic information, providing more personalized treatment optionstwenty three. These machine learning methods not only improve diagnostic efficiency for acute mastitis, but also help promote personalized medicine, making clinical decisions more accuratetwenty four.
In this study, logistic regression models were also applied to fit the data, and the results highlighted several important variables that are significantly related to the development of acute mastitis. These include age, nipple cracks, CRP levels, NE counts, and WBC counts. Using these insights, a comprehensive machine learning-based prediction model was developed to demonstrate its high accuracy and accuracy. Various machine learning algorithms, including logistic regression, were used to train and validate this model, and ultimately provided a powerful tool to predict the probability of acute mastitis in lactating women.
The predictive models described in this study showed excellent ability to distinguish. It could effectively separate patients suffering from acute mastitis from healthy, breastfed patients. In particular, he was proficient at identifying high-risk individuals. This skill is very clinically important. Provides important support for decisions made in a clinical setting. Compared to previous studies, the model in this study showed better sensitivity and Auroc5. This highlights the promise of machine learning technology. Improves the assessment of clinical risks of acute mastitistwenty five.
The findings of this study emerged as a significantly important factor associated with the onset of acute mastitis, with clear links observed in both onset and progression of the disease. Acute mastitis affects various proportions (2%-30%) of breastfeeding women worldwide, and most often occurs in the first 3 weeks after birth26. Young women who have recently given birth, especially those who start breastfeeding immediately after giving birth, are at a higher risk of developing acute mastitis. Previous studies have pointed out that the average age of women with lactating mastitis or breast abscess is 29 years old, reinforcing the notion that younger women are prone to this health problem.5.
Nipple damage was found to increase the risk of acute mastitis. Such damage includes cracks. It also includes abnormalities. These can be inversion or deformation. These issues can lead to blockage of milk production ducts. It can also interfere with the normal flow of milk27,28. The findings of this study confirm previous findings. They emphasize the importance of keeping your nipples in good condition. This is essential to prevent the onset of acute mastitis.
CRP levels serve as an important indicator of inflammation. They are recognized in relation to the intensity of acute mastitis. In cases of mastitis, an increase in CRP levels is usually observed. This elevation occurs in both serum and breast milk. It shows the extent of inflammation affecting the whole body29,30. A study in patients with idiopathic granulomatitis (IGM) revealed remarkable results. Serum CRP and interleukin-6 (IL-6) levels were significantly increased in patients experiencing severe IgM. This was in contrast to patients with fewer symptoms of the disease.30,31. These results highlight the role of CRP. It acts as a marker of inflammation. This is related to both diagnosis and handling of mastitis cases.
Neutrophils (NEs) are extremely important in the inflammatory response to mastitis. They act as the main cells that fight pathogens. One such pathogen is Escherichia coli32. Changes in neutrophil DNA methylation have been observed in cows with mastitis. These changes are associated with shifts in gene expression and regulation of microRNAs. This can increase the onset of mastitis33. Additionally, proinflammatory cytokines play a role. An example is tumor necrosis factor alpha (TNF-alpha). This cytokine is produced by macrophages in the mammary gland. It helps to bring neutrophils to the site of infection. Furthermore, molecules like interleukin-8 (IL-8) increase neutrophil arrival34. Nitric oxide (NO) also affects neutrophil function. May control movement and activity during mastitis35.
WBCs, particularly polynuclear neutrophils (PMNs), are important in the immune response during mastitis. They move from the blood to the breast. Their role is to fight infections36. An inflammatory response is necessary to remove pathogens, but too much inflammation can damage tissue. This underscores the importance of properly controlling the immune response. Such control is necessary to avoid further harm to breast tissue37. This process is strictly controlled by both the innate and adaptive immune systems. T and B lymphocytes become active when they fill antigen presenting cells. It also responds to other immune signals38.
The predictive model developed in this study provides significant clinical significance and allows for the identification of women at high risk for acute mastitis. This allows for the timely implementation of interventions and preventive measures, reducing the incidence and severity of mastitis and improving patient outcomes. The model also provides new insights into the pathogenesis and pathophysiological mechanisms of acute mastitis and points to the pathway to future research.
However, clinical application of the model is constrained by the timing of biomarker measurements (e.g., CRP). This was performed only during routine postnatal examinations and during diagnosis. This may interfere with our ability to effectively predict risk before the onset of clinical symptoms. If these biomarkers only show significant changes after symptoms develop, the early prediction ability of the model is limited. Future research should investigate changes in these biomarkers prior to the onset of symptoms to enhance the early prediction ability of the model.
The findings of this study are encouraging, but it is important to acknowledge certain limitations. The first limit concerns the source of the data. Data were collected from only one healthcare facility. This fact can limit the ability to apply the results to other areas or groups of people.
Furthermore, the design of this study did not consider any risk factors associated with acute mastitis. Subsequent research efforts should investigate additional potential risk factors. These include genetic trends, exposure to environmental factors, and lifestyle choices. As a result, the developed predictive models may not cover all variables that affect disease initiation. Therefore, models need to be validated in larger and more diverse populations. This step is important to assess how widely applicable the model is.
Furthermore, subsequent research efforts should investigate additional potential risk factors. These include genetic trends, exposure to environmental factors, and lifestyle choices. Such searches can improve and improve the predictive power of the model.
