Background: Machine learning models learn about general behavior from data by finding the relationships between features. Our purpose was to develop a predictive model to identify and predict which subset of colorectal cancer patients are more likely to experience chemotherapy-induced toxicity and to determine the specific attributes that influence the presence of treatment-related side effects. Methods: The predictor was general toxicity, and for the construction of our data training, we selected 95 characteristics that represent the health state of 74 patients prior to their first round of chemotherapy. After the data were processed, Random Forest models were trained to offer an optimal balance between accuracy and interpretability. Results: We constructed a machine learning predictor with an emphasis on assessing the importance of numerical and categorical variables in relation to toxicity. Conclusions: The incorporation of artificial intelligence in personalizing colorectal cancer management by anticipating and overseeing toxicities more effectively illustrates a pivotal shift towards more personalized and precise medical care.

Background and Objectives: Cancer pain continues to be a major clinical problem nowadays. This study aims to evaluate the World Health Organization (WHO) analgesic ladder effectiveness in patients with colorectal cancer and develop machine learning models to predict treatment response for precision pain management. Materials and Methods: In a retrospective observational study, a total of 107 oncological patients were analyzed, with a detailed subgroup analysis of 42 patients with colorectal cancer, hospitalized between July and September in 2022. The pain assessment used numerical rating scales at baseline and 2–3 weeks follow-up. Clinical variables included demographics, disease staging, metastatic patterns, analgesic progression, and medication usage. Machine learning algorithms (e.g., Random Forest, CatBoost, XGBoost, and Neural Network) were used to predict pain reduction outcomes. The UMAP dimensionality reduction and clustering identified the patient phenotypes. Results: Statistical analyses included descriptive methods, Chi-square and Mann–Whitney tests, and the models’ performance was evaluated by AUC. Among patients with colorectal cancer, 73.8% achieved clinically pain improvement, with a mean reduction of 2.62 points and median improvement of 3.00 points. The metastatic site significantly affected outcomes: visceral metastases patients showed median improvement of 3.00 points with high variability, patients with bone metastases demonstrated heterogeneous responses (range: −2.00 to +8.00 points), while non-metastatic patients exhibited consistent improvement. Random Forest achieved optimal predictive performance (AUC: 0.9167), identifying the baseline pain score, bone metastases, Fentanyl usage, anticonvulsants, and antispasmodics as key predictive features. The clustering analysis revealed two distinct phenotypes, requiring different analgesic intensities. Conclusions: This study validates the WHO analgesic ladder effectiveness while demonstrating superior outcomes in patients with colorectal cancer. The machine learning models successfully predict the treatment response with excellent discriminative ability, supporting precision medicine implementation in cancer pain management.

Background/Objectives: Glioblastomas (GBMs) and brain metastases (BMs) are both frequent brain lesions. Distinguishing between them is crucial for suitable therapeutic and follow-up decisions, but this distinction is difficult to achieve, as it includes clinical, radiological and histopathological correlation. However, non-invasive AI examination of conventional and advanced MRI techniques can overcome this issue. Methods: We retrospectively selected 78 patients with confirmed GBM (39) and single BM (39), with conventional MRI investigations, consisting of T2W FLAIR and CE T1W acquisitions. The MRI images (DICOM) were evaluated by an AI segmentation tool, comparatively evaluating tumor heterogeneity and peripheral edema. Results: We found that GBMs are less edematous than BMs (p = 0.04) but have more internal necrosis (p = 0.002). Of the BM primary cancer molecular subtypes, NSCCL showed the highest grade of edema (p = 0.01). Compared with the ellipsoidal method of volume calculation, the AI machine obtained greater values when measuring lesions of the occipital and temporal lobes (p = 0.01). Conclusions: Although extremely useful in radiomics analysis, automated segmentation applied alone could effectively differentiate GBM and BM on a conventional MRI, calculating the ratio between their variable components (solid, necrotic and peripheral edema). Other studies applied to a broader set of participants are necessary to further evaluate the efficacy of automated segmentation.

Background/Objectives: Glioblastoma (GBM) and brain metastases (BMs) are both frequent brain lesions. Their distinction is crucial for suitable therapeutic and follow-up decisions, but difficult to achieve, as it includes clinical, radiological and histopathological correlation. However, non-invasive AI examination of conventional and advanced MRI techniques can overcome this issue. Methods: We retrospectively selected 78 patients with confirmed GBM (39) and single BM (39), with conventional MRI investigations, consisting of T2W FLAIR and CE T1W acquisitions. The MRI images (DICOM) have been evaluated by an AI segmentation tool, comparatively evaluating the tumor heterogeneity and peripheric edema. Results: We found that GBM are less edematous than BM (p=0.04), but have more internal necrosis (p=0.002). Of the BM primary cancer molecular subtypes, NSCCL showed the higher grade of edema (p=0.01). Compared to the ellipsoidal method of volume calculation, the AI machine obtained greater values in measuring lesions of the occipital and temporal lobes (p=0.01). Conclusions: Although extremely useful in radiomics analysis, automated segmentation applied alone could effectively differentiate GBM and BM on conventional MRI, calculating the ratio between their variable components (solid, necrotic and peripheric edema). Other studies applied to a broader set of participants are necessary to further evaluate the efficacy of automated segmentation.

Pediatric acute-onset neuropsychiatric presentations occurring in the context of prior streptococcal exposure remain clinically important but diagnostically inconclusive, particularly at the interface between Pediatric Acute-onset Neuropsychiatric Syndrome (PANS) and Pediatric Autoimmune Neuropsychiatric Disorders Associated with Streptococcal Infections (PANDAS). This cross-sectional observational study examined whether serological, psychometric, and electroencephalographic findings converged within a clinically referred pediatric psychiatric cohort at the index assessment. Children and adolescents evaluated for neuropsychiatric symptoms in the context of caregiver- or record-supported history suggestive of prior streptococcal exposure were recruited over a 12-month period through inpatient and outpatient child psychiatric services. Out of 154 screened cases, 96 with analyzable index assessment data were retained and stratified by antistreptolysin O (ASO) status. Symptom burden was quantified using the Pediatric Acute-onset Neuropsychiatric Syndrome 31-Item Symptom Rating Scale (PANS-31) which was treated primarily as a continuous dimensional outcome and examined in relation to ASO titers, time since the last reported streptococcal infection, electroencephalography (EEG) findings, and selected developmental and clinical-history variables. Higher ASO values were associated with greater PANS-31 symptom burden, whereas a shorter interval since the last reported streptococcal infection was associated with both higher ASO titers and higher symptom scores. PANS-31 showed good total scale internal consistency and meaningful domain-level convergence with age-appropriate Child Symptom Inventory-4 (CSI-4) and Adolescent Symptom Inventory-4 (ASI-4) domains. Exploratory within-cohort stratification using an operational PANS-31 threshold and alternative post hoc thresholds suggested that higher symptom burden subgroups were largely confined to the ASO-positive stratum; however, these threshold-based patterns were interpreted descriptively and not as diagnostic performance or predictive classification. These findings do not support a disease-specific biomarker model, but suggest that higher anti-streptococcal serology, more recent streptococcal exposure, and greater neuropsychiatric burden may cluster as a clinically interpretable association pattern in a selected real-world psychiatric sample.

Background/Objectives: Glioblastoma (GB) remains the most prevalent primary malignant brain tumor in adults, characterized by its aggressive nature and poor prognosis. The present study endeavored to contribute to the development of advanced computational tools for neuro-oncology by integrating artificial intelligence (AI)-based segmentation and multi-model machine learning (ML) approaches. Methods: A retrospective analysis was conducted on patients with GB. AI-driven algorithms were utilized to perform volumetric segmentation of GB. These quantitative metrics were subsequently integrated into a multi-model ML framework to analyze correlations with patient survival and evaluate the predictive accuracy of the resulting models. Results: A total of 79 patients were ultimately included in the study after meeting all eligibility criteria. The results showed that larger GB tumors were associated with shorter post-treatment survival. Necrotic patterns within GB tumors impacted patient survival rates and response to therapy. Quantitative volumetric analysis of tumor enhancement, shape features, and morphological metrics were associated with patient outcomes. The Neural Network remained the top ML model performer overall for discrimination, but the Random Forest model also showed strong practical performance. Conclusions: As a summary, our study contributes to the development of advanced computational tools for neuro-oncology by integrating AI-based segmentation and multi-model ML approaches, and the results highlight the importance of imaging biomarkers in understanding GB prognosis.