Sofia Spano
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Sofia Spano has a total of 21 co-authors. Here are the top co-authors by number of shared publications:
- Maurizio Cecconi
Department of Biomedical Sciences, Humanitas University, Milan, Italy
3 shared publications - Antonio Voza
16Department of Emergency Medicine, IRCCS Humanitas Research Hospital, Rozzano, Milan, Italy
2 shared publications - Elena Costantini
Department of Anesthesia and Intensive Care, IRCCS Humanitas Research Hospital, Via Manzoni 56, Rozzano, 20089 Milan, Italy
2 shared publications - Massimiliano Greco
Department of Biomedical Sciences, Humanitas University, Pieve Emanuele, Italy
2 shared publications - Romina Aceto
Department of Anesthesia and Intensive Care, IRCCS Humanitas Research Hospital, Via Manzoni 56, Rozzano, 20089 Milan, Italy
2 shared publications - Alberto Molteni
General Surgery Unit, Ospedale Eugenio Morelli, 23035 Sondalo, Italy
1 shared publication
Latest Publications
- Open Access
- 1 February 2023
- Published by MDPI AG in Algorithms
- Vol. 16 (2) , 76
Machine Learning for Early Outcome Prediction in Septic Patients in the Emergency Department
Background: Sepsis is one of the major causes of in-hospital death, and is frequent in patients presenting to the emergency department (ED). Early identification of high-risk septic patients is critical. Machine learning (ML) techniques have been proposed for identification and prognostication of ED septic patients, but these models often lack pre-hospital data and lack validation against early sepsis identification scores (such as qSOFA) and scores for critically ill patients (SOFA, APACHE II). Methods We conducted an electronic health record (EHR) study to test whether interpretable and scalable ML models predict mortality in septic ED patients and compared their performance with clinical scores. Consecutive adult septic patients admitted to ED over 18 months were included. We built ML models, ranging from a simple-classifier model, to unbalanced and balanced logistic regression, and random forest, and compared their performance to qSOFA, SOFA, and APACHE II scores. Results: We included 425 sepsis patients after screening 38,500 EHR for sepsis criteria. Overall mortality was 15.2% and peaked in patients coming from retirement homes (38%). Random forest, like balanced (0.811) and unbalanced logistic regression (0.863), identified patients at risk of mortality (0.813). All ML models outperformed qSOFA, APACHE II, and SOFA scores. Age, mean arterial pressure, and serum sodium were major mortality predictors. Conclusions: We confirmed that random forest models outperform previous models, including qSOFA, SOFA, and APACHE II, in identifying septic patients at higher mortality risk, while maintaining good interpretability. Machine learning models may gain further adoption in the future with increasing diffusion and granularity of EHR data, yielding the advantage of increased scalability compared to standard statistical techniques.
- 1 February 2023
- Published by MDPI AG in Algorithms
- Vol. 16 (2) , 76
Open AccessMachine Learning for Early Outcome Prediction in Septic Patients in the Emergency Department
Background: Sepsis is one of the major causes of in-hospital death, and is frequent in patients presenting to the emergency department (ED). Early identification of high-risk septic patients is critical. Machine learning (ML) techniques have been proposed for identification and prognostication of ED septic patients, but these models often lack pre-hospital data and lack validation against early sepsis identification scores (such as qSOFA) and scores for critically ill patients (SOFA, APACHE II). Methods We conducted an electronic health record (EHR) study to test whether interpretable and scalable ML models predict mortality in septic ED patients and compared their performance with clinical scores. Consecutive adult septic patients admitted to ED over 18 months were included. We built ML models, ranging from a simple-classifier model, to unbalanced and balanced logistic regression, and random forest, and compared their performance to qSOFA, SOFA, and APACHE II scores. Results: We included 425 sepsis patients after screening 38,500 EHR for sepsis criteria. Overall mortality was 15.2% and peaked in patients coming from retirement homes (38%). Random forest, like balanced (0.811) and unbalanced logistic regression (0.863), identified patients at risk of mortality (0.813). All ML models outperformed qSOFA, APACHE II, and SOFA scores. Age, mean arterial pressure, and serum sodium were major mortality predictors. Conclusions: We confirmed that random forest models outperform previous models, including qSOFA, SOFA, and APACHE II, in identifying septic patients at higher mortality risk, while maintaining good interpretability. Machine learning models may gain further adoption in the future with increasing diffusion and granularity of EHR data, yielding the advantage of increased scalability compared to standard statistical techniques.
- Open Access
- 15 January 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (2) , 317
Kinetics of the Cell Cycle Arrest Biomarkers (TIMP2 and IGFBP7) for the Diagnosis of Acute Kidney Injury in Critically Ill COVID-19 Patients
Background: Acute kidney injury (AKI) is highly prevalent in critical COVID-19 patients. The diagnosis and staging of AKI are based on serum creatinine (sCr) and urinary output criteria, with limitations in the functional markers. New cell-cycle arrest biomarkers [TIMP2]*[IGFBP7] have been proposed for early detection of AKI, but their role in critically ill COVID-19 patients is poorly understood. Methods: We conducted an observational study to assess the performance of [TIMP2]*[IGFBP7] for the detection of AKI in critical COVID-19 patients admitted to our intensive care unit (ICU). We sampled urinary [TIMP2]*[IGFBP7] levels at ICU admission, 12 h, 24 h, and 48 h, and compared the results to the development of AKI, as well as baseline and laboratory data. Results: Forty-one patients were enrolled. The median age was 66 years [57–72] and most were males (85%). Thirteen patients (31.7%) developed no/mild stage AKI, 19 patients (46.3%) moderate AKI, and nine patients (22.0%) severe AKI. The ICU mortality was 29.3%. sCr levels in the Emergency Department or at ICU admission were not significantly different according to AKI stage. [TIMP-2]*[IGFBP-7] urinary levels were elevated in severe AKI at 12 h after ICU admission, but not at ICU admission or 24 h or 48 h after ICU admission. Conclusion: Urinary biomarkers [TIMP-2]*[IGFBP-7] were generally increased in this population with a high prevalence of AKI, and were higher in patients with severe AKI measured at 12 h from ICU admission. Further studies are needed to evaluate the best timing of these biomarkers in this population.
- 15 January 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (2) , 317
Open AccessKinetics of the Cell Cycle Arrest Biomarkers (TIMP2 and IGFBP7) for the Diagnosis of Acute Kidney Injury in Critically Ill COVID-19 Patients
Background: Acute kidney injury (AKI) is highly prevalent in critical COVID-19 patients. The diagnosis and staging of AKI are based on serum creatinine (sCr) and urinary output criteria, with limitations in the functional markers. New cell-cycle arrest biomarkers [TIMP2]*[IGFBP7] have been proposed for early detection of AKI, but their role in critically ill COVID-19 patients is poorly understood. Methods: We conducted an observational study to assess the performance of [TIMP2]*[IGFBP7] for the detection of AKI in critical COVID-19 patients admitted to our intensive care unit (ICU). We sampled urinary [TIMP2]*[IGFBP7] levels at ICU admission, 12 h, 24 h, and 48 h, and compared the results to the development of AKI, as well as baseline and laboratory data. Results: Forty-one patients were enrolled. The median age was 66 years [57–72] and most were males (85%). Thirteen patients (31.7%) developed no/mild stage AKI, 19 patients (46.3%) moderate AKI, and nine patients (22.0%) severe AKI. The ICU mortality was 29.3%. sCr levels in the Emergency Department or at ICU admission were not significantly different according to AKI stage. [TIMP-2]*[IGFBP-7] urinary levels were elevated in severe AKI at 12 h after ICU admission, but not at ICU admission or 24 h or 48 h after ICU admission. Conclusion: Urinary biomarkers [TIMP-2]*[IGFBP-7] were generally increased in this population with a high prevalence of AKI, and were higher in patients with severe AKI measured at 12 h from ICU admission. Further studies are needed to evaluate the best timing of these biomarkers in this population.
- Open Access
- 12 October 2022
- Published by MDPI AG in Journal of Personalized Medicine
- Vol. 12 (10) , 1705
The Cardiac Power Index during Abdominal Open Aortic Surgery: Intraoperative Insights into the Cardiac Performance—A Retrospective Observational Analysis
Background: The Cardiac Power Index (CPI) measures the rate of energy output generated by the heart and correlates this with in-hospital mortality due to cardiogenic shock. In open aortic surgery, both aortic clamping and unclamping expose the heart to abrupt variations of the left ventricle afterload, preload, and contractility, with possible hemodynamic impairment. We investigated how aortic-cross clamping (Ao-XC) and unclamping (Ao-UC) procedures affect the CPI during open aortic surgery. Methods: We retrospectively analyzed our surgical database of 67 patients submitted to open surgical aortic repair at Humanitas Research Hospital, Milan. Patients were monitored by an EV1000-FloTrac SystemTM (Edwards Lifescience, Irvine, CA, USA) beyond the standard intra-operative hemodynamic monitoring. The primary outcome was the variation of basal CPI after aortic clamping and unclamping. Secondary outcomes were variations of the cardiac index (CI), mean arterial pressure (MAP), heart rate, and lactate during aortic clamping and after unclamping. The CPI was computed as: (CI × MAP)/451. Results: The CPI changed significantly after aortic unclamping. CPI: basal = 0.39 ± 0.1 W/m2, after Ao-XC = 0.39 ± 0.1 W/m2, and after Ao-UC = 0.44 ± 0.2 W/m2, p < 0.05. The CI changed during both cross-clamping and unclamping (p < 0.0001), whilst the MAP and heart rate did not during any phase of the surgery. Five subjects (8.3%) needed inotropic support after cross-clamping. Their basal CPI was lower than the general population: 0.31 ± 0.11 W/m2 vs. 0.39 ± 0.1 W/m2. Conclusions: The CPI describes the adaptation of the cardiac function to the changes in preload, contractility, and afterload occurring during aortic cross-clamping and unclamping. It may be used to explore the cardiac performance in real-time and predict cardiac impairment in the intraoperative period in a minimally invasive way, similar to ventriculo-arterial coupling parameters.
- 12 October 2022
- Published by MDPI AG in Journal of Personalized Medicine
- Vol. 12 (10) , 1705
Open AccessThe Cardiac Power Index during Abdominal Open Aortic Surgery: Intraoperative Insights into the Cardiac Performance—A Retrospective Observational Analysis
Background: The Cardiac Power Index (CPI) measures the rate of energy output generated by the heart and correlates this with in-hospital mortality due to cardiogenic shock. In open aortic surgery, both aortic clamping and unclamping expose the heart to abrupt variations of the left ventricle afterload, preload, and contractility, with possible hemodynamic impairment. We investigated how aortic-cross clamping (Ao-XC) and unclamping (Ao-UC) procedures affect the CPI during open aortic surgery. Methods: We retrospectively analyzed our surgical database of 67 patients submitted to open surgical aortic repair at Humanitas Research Hospital, Milan. Patients were monitored by an EV1000-FloTrac SystemTM (Edwards Lifescience, Irvine, CA, USA) beyond the standard intra-operative hemodynamic monitoring. The primary outcome was the variation of basal CPI after aortic clamping and unclamping. Secondary outcomes were variations of the cardiac index (CI), mean arterial pressure (MAP), heart rate, and lactate during aortic clamping and after unclamping. The CPI was computed as: (CI × MAP)/451. Results: The CPI changed significantly after aortic unclamping. CPI: basal = 0.39 ± 0.1 W/m2, after Ao-XC = 0.39 ± 0.1 W/m2, and after Ao-UC = 0.44 ± 0.2 W/m2, p < 0.05. The CI changed during both cross-clamping and unclamping (p < 0.0001), whilst the MAP and heart rate did not during any phase of the surgery. Five subjects (8.3%) needed inotropic support after cross-clamping. Their basal CPI was lower than the general population: 0.31 ± 0.11 W/m2 vs. 0.39 ± 0.1 W/m2. Conclusions: The CPI describes the adaptation of the cardiac function to the changes in preload, contractility, and afterload occurring during aortic cross-clamping and unclamping. It may be used to explore the cardiac performance in real-time and predict cardiac impairment in the intraoperative period in a minimally invasive way, similar to ventriculo-arterial coupling parameters.