Multi-Institutional Validation of a Mammography-Based Breast Cancer Risk Model.

PURPOSE: Accurate risk assessment is essential for the success of population screening programs in breast cancer. Models with high sensitivity and specificity would enable programs to target more elaborate screening efforts to high-risk populations, while minimizing overtreatment for the rest. Artificial intelligence (AI)-based risk models have demonstrated a significant advance over risk models used today in clinical practice. However, the responsible deployment of novel AI requires careful validation across diverse populations. To this end, we validate our AI-based model, Mirai, across globally diverse screening populations. METHODS: We collected screening mammograms and pathology-confirmed breast cancer outcomes from Massachusetts General Hospital, USA; Novant, USA; Emory, USA; Maccabi-Assuta, Israel; Karolinska, Sweden; Chang Gung Memorial Hospital, Taiwan; and Barretos, Brazil. We evaluated Uno's concordance index for Mirai in predicting risk of breast cancer at one to five years from the mammogram. RESULTS: A total of 128,793 mammograms from 62,185 patients were collected across the seven sites, of which 3,815 were followed by a cancer diagnosis within 5 years. Mirai obtained concordance indices of 0.75 (95% CI, 0.72 to 0.78), 0.75 (95% CI, 0.70 to 0.80), 0.77 (95% CI, 0.75 to 0.79), 0.77 (95% CI, 0.73 to 0.81), 0.81 (95% CI, 0.79 to 0.82), 0.79 (95% CI, 0.76 to 0.83), and 0.84 (95% CI, 0.81 to 0.88) at Massachusetts General Hospital, Novant, Emory, Maccabi-Assuta, Karolinska, Chang Gung Memorial Hospital, and Barretos, respectively. CONCLUSION: Mirai, a mammography-based risk model, maintained its accuracy across globally diverse test sets from seven hospitals across five countries. This is the broadest validation to date of an AI-based breast cancer model and suggests that the technology can offer broad and equitable improvements in care.

Eosinopenia as predictor of infection in patients admitted to an internal medicine ward: a cross-sectional study.

BACKGROUND: The identification of infection in an internal medicine ward is crucial but not always straightforward. Eosinopenia has been proposed as a marker of infection, but specific cutoffs for prediction are not established yet. We aim to assess whether there is difference in eosinophil count between infected and noninfected patients and, if so, the best cutoffs to differentiate them. METHODS: Cross-sectional, observational study with analysis of all patients admitted to an Internal Medicine Department during 2 consecutive months. Clinical, laboratory and imaging data were analyzed. Infection at hospital admission was defined in the presence of either a microbiological isolation or suggestive clinical, laboratory, and/or imaging findings. Use of antibiotics in the 8 days before hospital admission, presence of immunosuppression, hematologic neoplasms, parasite, or fungal infections were exclusion criteria. In case of multiple hospital admissions, only the first admission was considered.Sensitivity and specificity values for eosinophils, leukocytes, neutrophils, and C-reactive protein were determined by receiver operating characteristic curve. Statistical analysis was performed with IBM SPSS Statistics® v25 and MedCalc Statistical Software® v19.2.3. RESULTS: A total of 323 hospitalization episodes were evaluated, each corresponding to a different patient. One hundred fifteen patients were excluded. A total of 208 patients were included, 62.0% (n = 129) of them infected at admission. Ten patients had multiple infections.Infected patients had fewer eosinophils than uninfected patients (15.8 ±â€Š42 vs 71.1 ±â€Š159 cell/mm3; P < .001). An eosinophil count at admission ≤69 cell/mm3 had a sensitivity of 89.1% and specificity of 54.4% (area under the curve 0.752; 95% confidence interval 0.682-0.822) for the presence of infection. Eosinophil count of >77 cells/mm3 had a negative likelihood ratio of 0.16. CONCLUSIONS: Eosinophil count was significantly lower in infected than in uninfected patients. The cutoff 69 cells/mm3 was the most accurate in predicting infection. Eosinophil count >77 cells/mm3 was a good predictor of absence of infection.