Machine learning based readmission and mortality prediction in heart failure patients.

This study intends to predict in-hospital and 6-month mortality, as well as 30-day and 90-day hospital readmission, using Machine Learning (ML) approach via conventional features. A total of 737 patients remained after applying the exclusion criteria to 1101 heart failure patients. Thirty-four conventional features were collected for each patient. First, the data were divided into train and test cohorts with a 70-30% ratio. Then train data were normalized using the Z-score method, and its mean and standard deviation were applied to the test data. Subsequently, Boruta, RFE, and MRMR feature selection methods were utilized to select more important features in the training set. In the next step, eight ML approaches were used for modeling. Next, hyperparameters were optimized using tenfold cross-validation and grid search in the train dataset. All model development steps (normalization, feature selection, and hyperparameter optimization) were performed on a train set without touching the hold-out test set. Then, bootstrapping was done 1000 times on the hold-out test data. Finally, the obtained results were evaluated using four metrics: area under the ROC curve (AUC), accuracy (ACC), specificity (SPE), and sensitivity (SEN). The RFE-LR (AUC: 0.91, ACC: 0.84, SPE: 0.84, SEN: 0.83) and Boruta-LR (AUC: 0.90, ACC: 0.85, SPE: 0.85, SEN: 0.83) models generated the best results in terms of in-hospital mortality. In terms of 30-day rehospitalization, Boruta-SVM (AUC: 0.73, ACC: 0.81, SPE: 0.85, SEN: 0.50) and MRMR-LR (AUC: 0.71, ACC: 0.68, SPE: 0.69, SEN: 0.63) models performed the best. The best model for 3-month rehospitalization was MRMR-KNN (AUC: 0.60, ACC: 0.63, SPE: 0.66, SEN: 0.53) and regarding 6-month mortality, the MRMR-LR (AUC: 0.61, ACC: 0.63, SPE: 0.44, SEN: 0.66) and MRMR-NB (AUC: 0.59, ACC: 0.61, SPE: 0.48, SEN: 0.63) models outperformed the others. Reliable models were developed in 30-day rehospitalization and in-hospital mortality using conventional features and ML techniques. Such models can effectively personalize treatment, decision-making, and wiser budget allocation. Obtained results in 3-month rehospitalization and 6-month mortality endpoints were not astonishing and further experiments with additional information are needed to fetch promising results in these endpoints.

Post-revascularization Ejection Fraction Prediction for Patients Undergoing Percutaneous Coronary Intervention Based on Myocardial Perfusion SPECT Imaging Radiomics: a Preliminary Machine Learning Study.

In this study, the ability of radiomics features extracted from myocardial perfusion imaging with SPECT (MPI-SPECT) was investigated for the prediction of ejection fraction (EF) post-percutaneous coronary intervention (PCI) treatment. A total of 52 patients who had undergone pre-PCI MPI-SPECT were enrolled in this study. After normalization of the images, features were extracted from the left ventricle, initially automatically segmented by k-means and active contour methods, and finally edited and approved by an expert radiologist. More than 1700 2D and 3D radiomics features were extracted from each patient's scan. A cross-combination of three feature selections and seven classifier methods was implemented. Three classes of no or dis-improvement (class 1), improved EF from 0 to 5% (class 2), and improved EF over 5% (class 3) were predicted by using tenfold cross-validation. Lastly, the models were evaluated based on accuracy, AUC, sensitivity, specificity, precision, and F-score. Neighborhood component analysis (NCA) selected the most predictive feature signatures, including Gabor, first-order, and NGTDM features. Among the classifiers, the best performance was achieved by the fine KNN classifier, which yielded mean accuracy, AUC, sensitivity, specificity, precision, and F-score of 0.84, 0.83, 0.75, 0.87, 0.78, and 0.76, respectively, in 100 iterations of classification, within the 52 patients with 10-fold cross-validation. The MPI-SPECT-based radiomic features are well suited for predicting post-revascularization EF and therefore provide a helpful approach for deciding on the most appropriate treatment.

Assessment of LV diastolic dysfunction in myocardial perfusion imaging: a correlative study with transthoracic echocardiography.

OBJECTIVE: Myocardial perfusion imaging (MPI) by gated single-photon emission computed tomography (SPECT) is a feasible method in the evaluation of left ventricular perfusion and function. The purpose of this study was to determine the threshold and grading of left ventricular (LV) diastolic dysfunction (LVDD) using gated SPECT MPI. METHODS: A total of 149 patients were recruited in the study. All of the patients underwent a standard 2-day stress/rest gated MPI study and transthoracic echocardiography within 2 weeks. The reconstructed rest-only images were analyzed by Cedar-Sinai's quantitative gated SPECT and the LV diastolic parameters, including peak filling rate (PFR), time to PFR (TTPF) and secondary PFR (PFR2) to PFR ratio were provided and compared to echocardiographic data. RESULTS: 68 (45.6%) and 81 (54.4%) of patients were categorized in LVDD-absent and LVDD-present groups on the basis of LVDD evidence in echocardiography, respectively. receiver-operating-characteristic analysis for PFR and TTPF was performed, resulting in diagnostic sensitivities of 70 and 57% and specificities of 60 and 75% for PFR <2.6 end-diastolic volumes (EDV)/s and TTPF>160.5 ms, respectively. Applying our previously used thresholds of <1.70 EDV/s for PFR, >208 ms for TTPF and >1 for PFR2/PFR, sensitivities and specificities of 9.9 and 96.6%, 9.9 and 95.6% and 13.8 and 88% were resulted, respectively. Grading of LVDD on the basis of MPI-obtained diastolic parameters showed considerable overlapping data by interquartile range. CONCLUSION: Gated SPECT MPI can be used as a highly specific means for detection of LV diastolic dysfunction when compared to echocardiography. However, grading of severity of diastolic heart failure appears to be impracticable.

Synthesis and characterization of Bombesin-superparamagnetic iron oxide nanoparticles as a targeted contrast agent for imaging of breast cancer using MRI.

The targeted delivery of superparamagnetic iron oxide nanoparticles (SPIONs) as a contrast agent may facilitate their accumulation in cancer cells and enhance the sensitivity of MR imaging. In this study, SPIONs coated with dextran (DSPIONs) were conjugated with bombesin (BBN) to produce a targeting contrast agent for detection of breast cancer using MRI. X-ray diffraction, transmission electron microscopy, and vibrating sample magnetometer analyses indicated the formation of dextran-coated superparamagnetic iron oxide nanoparticles with an average size of 6.0 ± 0.5 nm. Fourier transform infrared spectroscopy confirmed the conjugation of the BBN with the DSPIONs. A stability study proved the high optical stability of DSPION-BBN in human blood serum. DSPION-BBN biocompatibility was confirmed by cytotoxicity evaluation. A binding study showed the targeting ability of DSPION-BBN to bind to T47D breast cancer cells overexpressing gastrin-releasing peptide (GRP) receptors. T2-weighted and T2*-weighted color map MR images were acquired. The MRI study indicated that the DSPION-BBN possessed good diagnostic ability as a GRP-specific contrast agent, with appropriate signal reduction in T2*-weighted color map MR images in mice with breast tumors.

Impact of metoclopramide on image quality in myocardial perfusion imaging.

BACKGROUND: The effectiveness of metoclopramide in reducing gastrointestinal-induced artifacts in myocardial perfusion imaging (MPI) is a subject of debate. We examined the significance of this pharmacological intervention in the quality of images obtained from MPI studies. PATIENTS AND METHODS: A total of 211 suspected or known cases with coronary artery disease routinely referred to our nuclear medicine department for MPI were randomly assigned to group A and group B. Group A (N=125) comprised patients who received 10 mg of metoclopramide orally after the injection of the radiotracer [technetium-99m-labeled methoxyisobutyl isonitril (99mTc-MIBI)] 1 h before image acquisition, and group B (N=86) comprised patients who did not receive any pharmacological intervention and were considered the control group. All the scans in each group were assessed in the rest phase of a routine 2-day protocol. The single-photon emission computerized tomography (SPECT) images were visually evaluated in terms of extracardiac activities and their effects on image quality by three nuclear medicine physicians, who were blinded to the details of the protocol. RESULTS: Of the 125 patients who had received metoclopramide, 16 (13%) had nonacceptable, 72 (57.6%) had acceptable (interpretable), and 37 (29.6%) had good image quality. The image quality in group B was nonacceptable in 10 (11.23%), acceptable in 48 (50.23%), and good in 28 (33.56%) patients. The overall interobserver agreement was good (κ: 0.6-0.9, P<0.05) among the three readers. CONCLUSION: There was no statistically significant difference in terms of MPI-SPECT image quality between patients who received metoclopramide and those in the control group. Metoclopramide, therefore, did not exert a remarkable effect on the quality of our MPI scans.

Comparison of (99m)Tc-labeled PR81 and its F(ab')2 fragments as radioimmunoscintigraphy agents for breast cancer imaging.

OBJECTIVE: We digested anti-MUC1 monoclonal antibody PR81 to produce F(ab')2 fragments. A comparison was performed between the two radiolabeled PR81 and F(ab')2 fragments for breast tumor imaging in a mouse model. METHODS: The optimum conditions for pepsin digestion of PR81 were investigated in terms of enzymes: antibody ratio, digestion time duration and preserved immunoreactivity of the produced fragments. The F(ab')2 fragments were labeled with Technetium-99m using HYNIC as a chelator and tricine as a co-ligand. The immunoreactivity of the complexes was assessed by radioimmunoassay using MCF7 cells. Biodistribution and imaging studies were performed in female BALB/c mice with breast tumor xenograft at 4, 8 and 24 h post-administration. The PR81 was labeled with technetium-99m in the same way for comparison. RESULTS: The optimum time duration for PR81 digestion was found to be 28 h at an enzyme:antibody weight ratio of 1:20 that resulted in 95.2 ± 4.7% purity. The labeling of intact PR81 and its F(ab')2 fragments were 87.6 ± 4.2 and 76.1 ± 3.3% after 1 h, respectively (p value <0.05). The percentage of immunoreactivity of F(ab')2 fragments and intact PR81 were 75.4 ± 2.1% and 85.7 ± 2.9%, respectively (p value <0.05). The biodistribution and imaging studies demonstrated localization of the fragments at 4 h post-administration with high sensitivity and specificity. CONCLUSION: The results showed that F(ab')2 fragment of PR81 is more suitable than intact PR81 for safer and more rapid detection of human breast cancer.