Using Machine Learning to Identify No-Show Telemedicine Encounters in a New York City Hospital

No-show visits are a serious problem for healthcare centers. It costs a major hospital over 15 million dollars annually. The goal of this paper was to build machine learning models to identify potential no-show telemedicine visits and to identify significant factors that affect no-show visits. 257,293 telemedicine sessions and 152,164 unique patients were identified in Mount Sinai Health System between March 2020 and December 2020. 5,124 (2%) of these sessions were no-show encounters. Extreme Gradient Boosting (XGB) with under-sampling was the best machine learning model to identify no-show visits using telemedicine service. The accuracy was 0.74, with an AUC score of 0.68. Patients with previous no-show encounters, non-White or non-Asian patients, and patients living in Bronx and Manhattan were all important factors for no-show encounters. Furthermore, providers' specialties in psychiatry and nutrition, and social workers were more susceptible to higher patient no-show rates.

Using EHR Data to Identify Social Determinants of Health Affecting Disparities in Cancer Survival

The aim of this pilot study was to identify social determinants of health (SDH) that affect disparities in cancer survival. A limited dataset was generated by querying electronic medical records (EHR) from an academic medical center in New York City between January 2003 and November 2020. Socio-demographic characteristics that affected survival in 22,096 cancer patients were analyzed using descriptive statistics and logistic regression analyses. Two subsets of adult patients were identified: patients who were deceased less than 1 year after diagnosis and patients who survived over 5 years after diagnosis. Percentage of individuals with short survival in Blacks and Whites was respectively 41.4% and 22.2% for lung cancer, 9.8% and 7.1% for colorectal cancer, 2.9% and 0.7% for breast cancer, 6.8% and 4.0% for multiple myeloma, and 1.4% and 0.8% for prostate cancer. Logistic regression identified SDH factors increasing likelihood of shorter survival that included older age, and being male, Black or Hispanic. We concluded that further analysis of a broader spectrum of SDH factors is warranted.

Using EHR data to compare survival disparities in major cancer types

e18517

Background: The aim of our study is to provide a comparative overview of survival disparities in 5 major cancer types: breast cancer, colorectal cancer, lung cancer, multiple myeloma and prostate cancer, within one health system in New York City. Using this approach, we aim at initial assessment of potential drivers of disparities in cancer survival, especially for those who deceased in a short amount of time after diagnosis. Methods: A limited dataset was generated from electronic medical records from Mount Sinai Health System for the five cancer types from January 2003 to November 2020. We extracted 2 subsets from the original dataset: patients who survived less than 1 year between diagnoses and death, and patients who have survived for over 5 years after diagnoses. We performed logistic regression to investigate the effect of demographical factors on patients’ duration of survival after cancer diagnosis. Results: There are 1280 patients who survived less than 1 year and 20,816 patients who survived over 5 years. Although there are less overall colorectal cancer and lung cancer diagnoses, there are significantly more patients who deceased within a short of period of time from these 2 cancers, comparing to the other 3 cancer types. Race was an important factor. There was significantly more proportions of black patients deceased in a short period of time than in a long time after diagnosis. Percentage of individuals with short survival in Blacks and Whites was respectively 41.4% and 22.2% for lung cancer, 9.8% and 7.1% for colorectal cancer, 2.9% and 0.7% for breast cancer, 6.8% and 4.0% for multiple myeloma, and 1.4% and 0.8% for prostate cancer. This disparity was observed in all five cancer types. When comparing within race, there were significantly higher proportions of black patients survived a short time than those survived a long time for all 5 cancer types. When comparing with other races, black patients who were diagnosed with lung cancer or breast cancer were more likely to be deceased in a short time after diagnosis compare to white patients. And Hispanic patients with lung cancer were also at risk of shorter survival length compared to white patients. Thus, promoting routine cancer screening is important in black and Hispanic communities. Conclusions: Logistic regression identified factors increasing likelihood of shorter survival that included older age, and being male, Black or Hispanic. We concluded that further analysis of broader spectrum of risk factors is warranted.

Identifying Determinants of Disparities in Lung Cancer Survival Rates from Electronic Health Record Data

The goal of this pilot study was to identify significant factors that affect disparities in lung cancer survival. A de-identified dataset was generated by querying electronic health records (EHR) from an academic medical center in New York City between January 2003 and November 2020. Socio-demographic characteristics, cancer stage, and genetic profile were analyzed using logistic regression. Two subsets of adult patients were identified: patients who were deceased less than 1 year after diagnosis and patients who survived over 5 years after diagnosis. Male, Black and Hispanic patients and those who were diagnosed in later stages were the people most susceptible to a shorter length of survival after cancer diagnoses. In addition, we identified three genetic oncodrivers (KRAS, EGFR and TP53) which were highly correlated with the length of survival after lung cancer diagnoses and their distribution was associated with race. We concluded that EHR data provide important insights on cancer survival disparities.

Autophagy-Sirtuin1(SIRT1) Alleviated the Coronary Atherosclerosis (AS)in Mice through Regulating the Proliferation and Migration of Endothelial Progenitor Cells (EPCs) via wnt/β-catenin/GSK3β Signaling Pathway

BACKGROUND AND PURPOSE

SIRT1 was associated with AS risk and EPCs were reported to participate in the endothelial repair in Coronary Atherosclerosis (CAS). In this study, we explored the role of SIRT1 in AS mice and also its modulation in EPCs.

METHODS AND MATERIALS

ApoE-/-mice were fed on high-fat and high-glucose diet to establish the AS animal model with the normally-raised C57BL/6 mice as a control group. SIRT1 activator, SRT 2104 was injected intravenously into 5 ApoE-/-mice and its inhibitor Nicotinamide was injected in tail in another 5 ApoE-/-mice. Weight changes were recorded. Blood samples were taken from posterior orbital venous plexus and were detected by automatic biochemical analyzer. HE staining displayed the pathological conditions while Immunohistochemistry (IHC) evaluated the CD34+/VEGFR2+ relative density in the aorta tissues. EPCs were isolated from bone marrow and verified using immunofluorescence staining (IFS). The modulatory mechanism of SIRT1 in EPCs were studied by using RT-PCR, MTT, Western Blot and colony formation, scratch methods.

RESULTS

SIRT1 activator negatively regulated the weight and TC, TG and LDL levels, alleviated the lesion conditions and decreased the CD34+/VEGFR2+ density compared to the AS control. In vitro, SIRT1 activator promoted the proliferation and migration of EPCs and activated wnt/β-catenin/GSK3β signaling pathway. SIRT1 activator also inhibited the autophagy biomarkers ATG1 and LC3II. Furthermore, inhibitor of autophagy promoted SIRT1 expression and induced EPC proliferation, migration and activated wnt/β-catenin/GSK3β pathway. The suppression of the wnt/β-catenin/GSK3β pathway inhibited SIRT1 expression in EPCs, attenuated the proliferation and migration and promoted autophagy of EPCs.

CONCLUSION

SIRT1 activation might be protective in AS mice through autophagy inhibition in EPCs via wnt/β-catenin/GSK3β signaling pathway.

Comparative Analysis of Patient Distress in Opioid Treatment Programs using Natural Language Processing

Psychiatric and medical disorders, social and family environment, and legal distress are important determinants of distress that impact the effectiveness of the treatment in opioid treatment program (OTP). This information is not routinely captured in electronic health record, but may be found in clinical notes. This study aims to explore the feasibility and effectiveness of natural language processing (NLP) strategy for identifying legal, social, mental and medical determinates of distress along with emotional pain rooted in family environment from clinical narratives of patients with opioid addiction, and then using this information to find its impact on OTP outcomes. Analysis in this study showed that mental and legal distress significantly impact the result of the treatment in OTP.

The Three Hundred–NIKA2 Sunyaev–Zeldovich Large Program twin samples: Synthetic clusters to support real observations

The simulation database of The Three Hundred Project has been used to pick synthetic clusters of galaxies with properties close to the observational targets of the NIKA2 camera Sunyaev–Zeldovich (SZ) Large Program. Cross–matching of cluster parameters such as mass and redshift of the cluster in the two databases has been implemented to generate the so–called twin samples for the Large Program. This SZ Large Program is observing a selection of galaxy clusters at intermediate and high redshift (0:5 < z < 0:9), covering one order of magnitude in mass. These are SZ–selected clusters from the Planck and Atacama Cosmology Telescope catalogs, wherein the selection is based on their integrated Compton parameter values, Y500: the value of the parameter within the characteristics radius R500.

The Three Hundred hydrodynamical simulations provide us with hundreds of clusters satisfying these redshift, mass, and Y500 requirements. In addition to the standard post-processing analysis of the simulation, mock observational maps are available mimicking X–ray, optical, gravitational lensing, radio, and SZ observations of galaxy clusters. The primary goal of employing the twin samples is to compare different cluster mass proxies from synthetic X–ray, SZ effect and optical maps (via the velocity dispersion of member galaxies and lensing κ-maps) of the clusters. Eventually, scaling laws between different mass proxies and the cluster mass will be cross–correlated to reduce the scatter on the inferred mass and the mass bias will be related to various physical parameters.

The Three Hundred project: Contrasting clusters galaxy density in hydrodynamical and dark matter only simulations

Cluster number count is a major cosmological probe for the next generation of cosmological large scale-structure surveys like the one expected from the Euclid satellite mission. Cosmological constraints will be mainly limited by the understanding of the selection function (SF), which characterize the probability of detecting a cluster of a given mass and redshift. The SF can be estimated by injecting realistic simulated clusters into the survey and re-applying the detection procedure. For this purpose we intend to use The Three Hundreds project, a 324 cluster sample simulated with full-physics hydrodynamical re-simulations. In this paper we concentrate on the study of the distribution of member galaxies in the cluster sample. First, we study possible resolution effects by comparing low and high resolution simulations. Finally, accounting for the latter we derive the density profiles of the member galaxies and discuss their evolution with cluster mass and redshift.

Use of Artificial Intelligence for Predicting COVID-19 Outcomes: A Scoping Review

During the COVID-19 pandemic, artificial intelligence has played an essential role in healthcare analytics. Scoping reviews have been shown to be instrumental for analyzing recent trends in specific research areas. This paper aimed at applying the scoping review methodology to analyze the papers that used artificial intelligence (AI) models to forecast COVID-19 outcomes. From the initial 1,057 articles on COVID-19, 19 articles satisfied inclusion/exclusion criteria. We found that the tree-based models were the most frequently used for extracting information from COVID-19 datasets. 25% of the papers used time series to transform and analyze their data. The largest number of articles were from the United States and China. The reviewed artificial intelligence methods were able to predict cases, death, mortality, and severity. AI tools can serve as powerful means for building predictive analytics during pandemics.

Latent COVID-19 Clusters in Patients with Opioid Misuse

The goal of this paper is to apply unsupervised machine learning techniques in order to discover latent clusters in patients who have opioid misuse and also undergone COVID-19 testing. Target dataset has been constructed based on COVID-19 testing results at Mount Sinai Health System and opioid treatment program (OTP) information from New York State Office of Addiction Service and Support (OASAS). The dataset was preprocessed using factor analysis for mixed data (FAMD) method and then K-means algorithm along with elbow method were used to determine the number of optimal clusters. Four patient clusters were identified among which the fourth cluster constituted the maximum percentage of positive COVID-19 test results (20%). Compared to the other clusters, this cluster has the highest percentage of African Americans. This cluster has also the highest mortality rate (16.52%), hospitalization rate after receiving the COVID-19 test result (72.17%, use of ventilator (7.83%) and ICU admission rate (47.83%). In addition, this cluster has the highest percentage of patients with at least one chronic disease (99.13%) and age-adjusted comorbidity score more than 1 (83.48%). Longer participation in OTP was associated with the highest morbidity and mortality from COVID-19.

Machine Learning Approaches for Early Prostate Cancer Prediction Based on Healthcare Utilization Patterns

The goal of this study was to build a machine learning model for early prostate cancer prediction based on healthcare utilization patterns. We examined the frequency and pattern changes of healthcare utilization in 2916 prostate cancer patients 3 years prior to their prostate cancer diagnoses and explored several supervised machine learning techniques to predict possible prostate cancer diagnosis. Analysis of patients' medical activities between 1 year and 2 years prior to their prostate cancer diagnoses using XGBoost model provided the best prediction accuracy with high F1 score (0.9) and AUC score (0.73). These pilot results indicated that application of machine learning to healthcare utilization patterns may result in early identification of prostate cancer diagnosis.

Molecular and phenotypic characteristics of 15q24 microdeletion in pediatric patients with developmental disorders

Chromosome 15q24 microdeletion is a rare genetic disorder characterized by development delay, facial dysmorphism, congenital malformations, and occasional autism spectrum disorder (ASD). In this study, we identified five cases of 15q24 microdeletion using multiplex ligation-dependent probe amplification (MLPA) technology in a cohort of patients with developmental delay and/or intellectual disability. Two of these five cases had deletions that overlapped with the previously defined 1.1 Mb region observed in most reported cases. Two cases had smaller deletions (< 0.57 Mb) in the 15q24.1 low copy repeat (LCR) B-C region. They presented significant neurobehavioral features, suggesting that this smaller interval is critical for core phenotypes of 15q24 microdeletion syndrome. One case had minimal homozygous deletion of less than 0.11 Mb in the 15q24.1 LCR B-C region, which contained CYP1A1 (cytochrome P450 family 1 subfamily A member 1) and EDC3 (enhancer of mRNA decapping 3) genes, resulting in poor immunity, severe laryngeal stridor, and lower limbs swelling. This study provides additional evidence of 15q24 microdeletion syndrome with genetic and clinical findings. The results will be of significance to pediatricians in their daily practice.

[Identification of serum peptide biomarker for ovarian cancer diagnosis by Clin-TOF-II-MS combined with magnetic beads technology]

Objective: To explore the serum cyclic polypeptide biomarkers for ovarian cancer diagnosis. Methods: A total of 54 patients with epithelial ovarian cancer confirmed by pathology in Cancer Hospital, Chinese Academy of Medical Sciences from March 2018 to September 2018 were selected as the study subjects, and 40 healthy women with normal examination results in the cancer screening center were selected as the control. All of the samples were randomly divided into training set and validation set at the ratio of 1∶1 with a random number. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) combined with magnetic bead technology was used for detecting peptide profiling in serum samples to screen significantly differently expressed peptides between ovarian cancer group and control group of the training set (score>5). Receiver operating characteristic (ROC) curve analysis was used to screen differential peptide peaks with area under curve (AUC) ≥0.8, sensitivity and specificity>90% in the training set and validation set. Liquid chromatography-mass spectrometry (LC-MS/MS) was further used to determine the composition of differentially expressed peptides. Results: By comparing the peptide profiles of the two groups, 102 differential peptide peaks were initially detected in the mass-to-charge ratio range of 1 000 to 10 000. ROC curve analysis showed that there were 42 differential peptide peaks with AUC ≥0.8 in both training set and validation set, 19 of which were highly expressed in ovarian cancer group, and 23 were lowly expressed. There were 15 different peptide peaks in highly expressed ovarian cancer group with sensitivity and specificity over 90%. The mass-to-charge ratios were 7 744.27, 5 913.41, 5 329.87, 4 634.21, 4 202.02, 3 879.26, 3 273.35, 3 253.79, 3 234.34, 2 950.33, 2 664.51, 2 018.38, 1 893.37, 1 498.69 and 1 287.55. There were 15 different peptide peaks in lowly expressed ovarian cancer group with sensitivity and specificity over 90%, the mass-to-charge ratios were 9 288.46, 7 759.77, 5 925.24, 4 652.77, 4 210.42, 3 887.02, 3 279.90, 3 240.82, 2 962.15, 2 932.70, 2 022.42, 1 897.16, 1 501.69, 1 337.38 and 1 290.13. No protein composition was identified in 15 different peptide peaks in lowly expressed ovarian cancer group. The two protein compositions identified in 15 different peptide peaks in highly expressed ovarian cancer group were recombinant serglycin (SRGN) and fibinogen alpha chain (FGA), the mass-to-charge ratios of which were 1 498.696 and 5 913.417, respectively. The sensitivity and specificity of the two proteins for ovarian cancer diagnosis were 100%, 100% and 90.9%, 100%, respectively. Conclusion: SRGN and FGA are highly expressed in the serum of ovarian cancer patients, which may be potential diagnostic markers for ovarian cancer.

[Association of GMF-β expression with Ki-67 and its significance in the prognostic evaluation of astrocytoma]

Objective: To clarify the correlation of the expression of glia maturation factor-β (GMF-β) with Ki-67 in astrocytoma, and to investigate the prognostic implications of combined detection of GMF-β and Ki-67. Methods: One hundred and forty human astrocytoma samples (WHO Ⅱ-Ⅳ grade) were collected at Southwest Hospital, Army Medical University (the Third Military Medical University), China from 2006 to 2009. Clinicopathological information and 3-year follow-up data were collected. Expression of GMF-β and Ki-67 was detected by single and double immunohistochemical staining, then the association of GMF-β expression with Ki-67 and its significance in prognostic evaluation of astrocytoma were statistically analyzed. Results: GMF-β expression in astrocytoma cells was correlated to both tumor grade and Ki-67 (both P<0.05); Kaplan-Meier survival analysis showed that GMF-β and Ki-67 expression were negatively correlated to the 3 year-survival rates, respectively (both P<0.01). Further analysis demonstrated that the two factors were co-influenced on survival, showing a trend of "GMF-β^low Ki-67^low>GMF-β^high Ki-67^low>GMF-β^low Ki-67^high>GMF-β^high Ki-67^high" in 3-year survival rate with significant intergroup differences (P<0.05, P<0.01). Conclusions: GMF-β expression is positively associated with Ki-67 in astrocytoma. Combined detection of GMF-β and Ki-67 can predict prognosis of patients with glioma.

Comparison of ACM and CLAMP for Entity Extraction in Clinical Notes

Rapid increase in adoption of electronic health records in health care institutions has motivated the use of entity extraction tools to extract meaningful information from clinical notes with unstructured and narrative style. This paper investigates the performance of two such tools in automatic entity extraction. In specific, this work focuses on automatic medication extraction performance of Amazon Comprehend Medical (ACM) and Clinical Language Annotation, Modeling and Processing (CLAMP) toolkit using 2014 i2b2 NLP challenge dataset and its annotated medical entities. Recall, precision and F-score are used to evaluate the performance of the tools.Clinical Relevance- Majority of data in electronic health records (EHRs) are in the form of free text that features a gold mine of patient's information. While computerized applications in healthcare institutions as well as clinical research leverage structured data. As a result, information hidden in clinical free texts needs to be extracted and formatted as a structured data. This paper evaluates the performance of ACM and CLAMP in automatic entity extraction. The evaluation results show that CLAMP achieves an F-score of 91%, in comparison to an 87% F-score by ACM.

Cell-free DNA from bile outperformed plasma as a potential alternative to tissue biopsy in biliary tract cancer

BACKGROUND

Biliary tract cancers (BTCs) are rare and highly heterogenous malignant neoplasms. Because obtaining BTC tissues is challenging, the purpose of this study was to explore the potential roles of bile as a liquid biopsy medium in patients with BTC.

PATIENTS AND METHODS

Sixty-nine consecutive patients with suspected BTC were prospectively enrolled in this study. Capture-based targeted sequencing was performed on tumor tissues, whole blood cells, plasma, and bile samples using a large panel consisting of 520 cancer-related genes.

RESULTS

Of the 28 patients enrolled in this cohort, tumor tissues were available in eight patients, and plasma and bile were available in 28 patients. Somatic mutations were detected in 100% (8/8), 71.4% (20/28), and 53.6% (15/28) of samples comprising tumor tissue DNA, bile cell-free DNA (cfDNA), and plasma cfDNA, respectively. Bile cfDNA showed a significantly higher maximum allele frequency than plasma cfDNA (P = 0.0032). There were 56.2% of somatic single-nucleotide variant (SNVs)/insertions and deletions (indels) shared between bile and plasma cfDNA. When considering the genetic profiles of tumor tissues as the gold standard, the by-variant sensitivity and positive predictive value for SNVs/indels in bile cfDNA positive for somatic mutations were both 95.5%. The overall concordance for SNVs/indels in bile was significantly higher than that in plasma (99.1% versus 78.3%, P < 0.0001). Moreover, the sensitivity of CA 19-9 combined with bile cfDNA achieved 96.4% in BTC diagnosis.

CONCLUSION

We demonstrated that bile cfDNA was superior to plasma cfDNA in the detection of tumor-related genomic alterations. Bile cfDNA as a minimally invasive liquid biopsy medium might be a supplemental approach to confirm BTC diagnosis.

[Predicting postoperative recurrence of stage Ⅰ-Ⅲ renal clear cell carcinoma based on preoperative CT radiomics feature nomogram]

OBJECTIVE

To explore the preoperative radiomics features (RFs) and construct a nomogram for predicting postoperative recurrence of stage Ⅰ-Ⅲ clear cell renal carcinoma (ccRCC).

METHODS

The clinicopathological data and preoperative enhanced CT images collected from 256 patients with ccRCC were used as the training dataset (175 patients) and test dataset (81 patients). The enhanced CT images of the tumor were segmented using ITK-SNAP software, and the RFs were extracted using the PyRadiomics computing platform. In the training dataset, the RFs were screened based on Lasso-CV algorithm, and the Rad_score was calculated. The Clinic factors were screened by univariate and multivariate logistic regression analysis of the clinical and pathological factors and CT characteristics. The Rad_score, Clinic、Rad_score + Clinic nomograms were constructed and verified using the test dataset. The performance, discrimination power and calibration of the nomograms were compared, and their clinical value was evaluated using decision curve analysis.

RESULTS

Six RFs were retained to calculate the Rad_score. The Clinic factors included Rad_score, KPS score, platelet, calcification and TNM clinical stage. In terms of discrimination, the Rad_score + Clinic nomogram showed better performance (AUC=0.84 for training set; AUC=0.85 for test set) than the Rad_score nomogram (AUC=0.78 for training set, P=0.029; AUC=0.77 for Test set, P=0.025) and Clinic nomogram (AUC=0.77 for training set, P=0.014; AUC=0.77 for test set, P=0.011). In terms of calibration, the P value for goodness of fit test of the Rad_score+Clinic nomogram was 0.065 for the training set and 0.628 for the test set. Decision curve analysis showed a greater clinical value of the Rad_score+Clinic nomogram with Rad_score than the Clinic nomogram without Rad_score.

CONCLUSION

The nomogram based on preoperative CT RFs has a high value for predicting postoperative recurrence of stage Ⅰ-Ⅲ ccRCC to facilitate individualized treatment of RCC.

[Risk factors of occurrence and treatment failure of peritoneal dialysis-associated polymicrobial peritonitis: a multicenter retrospective study]

OBJECTIVE

To determine the risk factors of occurrence and treatment failure of peritoneal dialysis associatedperitonitis (PDAP) due to polymicrobial infections.

METHODS

We retrospectively collected the clinical data of patients with PDAP from the peritoneal dialysis (PD) centers in 4 general hospitals in Jilin Province from 2013 to 2019. The patients were divided, according to the results of peritoneal dialysate culture, into polymicrobial PDAP group and control group for comparison of the clinical data, treatment outcomes, and long-term prognosis. The independent risk factors of the occurrence and treatment failure of polymicrobial PDAP were explored using multivariate regression analysis.

RESULTS

We recruited a total of 625 patients from the 4 PD centers, among whom 1085 episodes of PDAP were recorded. Polymicrobial PDAP accounted for 7.6% of the total PDAP episodes, and this proportion increased from 5.3% in 2013-2016 to 9.4% in 2017-2019 (P= 0.012). Compared with the control group, polymicrobial PDAP group had higher proportions of elderly patients and patients with refractory PDAP, with greater white blood cell counts in the first-day dialysate and longer course of antibiotic treatment (P < 0.05). The risk of catheter removal and treatment failure (catheter removal or PDAP-related death) in polymicrobial PDAP group was 2.972 times (OR=2.972, 95% CI: 1.634-5.407, P < 0.001) and 2.692 times (OR=2.692, 95% CI: 1.578-4.591, P < 0.001) that in the control group, respectively. The risk of withdrawal from PD (technical failure + all-cause death) was 1.5- fold higher in polymicrobial PDAP group than that in the control group (OR=1.500, 95% CI: 1.085-2.074, P=0.014). Elderly patients (>65 years) had a 1.937-fold higher risk of experiencing polymicrobial PDAP than younger patients (OR=1.937, 95% CI: 1.207-3.109, P= 0.006). Diabetes mellitus (OR=5.554, 95% CI: 1.021-30.201, P=0.047), mixed fungal infeciton (OR=343.687, 95% CI: 21.554- 5480.144, P < 0.001), and Pseudomonas aeruginosa infection (OR=11.518, 95% CI: 1.632 to 81.310, P=0.014) were associated with increased risks of treatment failure by 4.554, 342.687 and 10.518 times, respectively.

CONCLUSION

The proportion of polymicrobial PDAP in the total PDAP cases tends to increase in recent years. Polymicrobial infection is an independent risk factor of both treatment failure and poor prognosis in patients with PDAP. An old age is an independent risk factor for polymicrobial PDAP, while diabetes mellitus and infections with mixed fungi or Pseudomonas aeruginosa are independent risk factors for treatment failure.

Impact of COVID-19 Pandemic on Use of Telemedicine Services in an Academic Medical Center

The COVID-19 pandemic changed the landscape of telehealth services. The goal of this paper was to identify demographic groups of patients who have used telemedicine services before and after the start of the pandemic, and to analyze how different demographic groups' telehealth usage patterns change throughout the course of the pandemic. A de-identified study dataset was generated by querying electronic health records at the Mount Sinai Health System to identify all patients. 129,625 patients were analyzed. Demographic shifts in patients seeking telemedicine service were identified. There was significant increase in the middle age and older population using telehealth services. During the pandemic use of telemedicine services was increased among male patients and racial minority patients. Furthermore, telehealth services had expanded to a broader spectrum of medical specialties.

[Treatment of lumbar degenerative diseases with recapping laminoplasty and nerve root canal's decompression preserving the continuity of supraspinous ligament]

Objective: To explore the clinical effect of lumbar discectomy and nerve root canal's enlargement preserving the continuity of supraspinous ligament in the treatment of lumbar degenerative disease. Methods: The data of patients with lumbar degenerative disease who underwent operation from 2016 to 2018 were analyzed retrospectively, and the patients were divided into two groups according to the different operation. The treatment group (17 cases) was treated with recapping laminoplasty, lumbar discectomy and nerve root canal's enlargement, and the control group (28 cases) was treated with total laminectomy, nerve root canal's enlargement, lumbar discectomy, interbody fusion and internal fixation (PLIF). All patients were followed up for 12 to 27 months (mean 17.8 months). Japanese Orthopaedic Association Scores(JOA) and visual analogue scale(VAS) of pain were used to evaluate the clinical effect before and after the operation, lumbar dynamical X-ray and Cobb angle were collecting for imaging evaluation, and the adjacent segment degeneration at the last follow-up was recorded. Results: There was no significant difference in preoperative JOA score, VAS score and Lumbar Cobb angle between the two groups (all P>0.05). The operation time in the treatment group was shorter than that in the control group, and the blood loss during operation in the treatment group was lower than that in the control group, the bed rest time of the treatment group after operation was shorter than that in the control group ((79±14) vs (118±17) min, (151±38) vs (324±70) ml and (3.4±0.7) vs (4.3±1.0) d,respectively; t=-8.508, -10.724, -3.244, all P<0.01). In addition, compared with the control group, the volume of postoperative drainage in the treatment group also decreased significantly (t=-5.637, P<0.01). There was no significant difference in JOA score between the two groups 1 year after the operation (P>0.05), but there was significant difference in VAS score between the two groups, the treatment group was better than the control group (P<0.05). Compared with the control group, the lumbar Cobb angle in the treatment group increased significantly one year after the operation (55.3°±3.2° vs 38.4°±6.2°, t=10.391, P<0.05). During the follow-up, no loosening or fracture of the implants was found in all patients. Conclusion: Treatment of lumbar degenerative diseases with recapping laminoplasty and nerve root canal's decompression preserving the continuity of supraspinous ligament by ultrasound osteotome has the same clinical effect as PLIF. It has the advantages of shortening operation time, less bleeding, better maintenance of lumbar lordosis after operation and reduction of adjacent segment degeneration.

Latent COVID-19 Clusters in Patients with Chronic Respiratory Conditions

The goal of this paper was to apply unsupervised machine learning techniques towards the discovery of latent COVID-19 clusters in patients with chronic lower respiratory diseases (CLRD). Patients who underwent testing for SARS-CoV-2 were identified from electronic medical records. The analytical dataset comprised 2,328 CLRD patients of whom 1,029 were tested COVID-19 positive. We used the factor analysis for mixed data method for preprocessing. It performed principle component analysis on numeric values and multiple correspondence analysis on categorical values which helped convert categorical data into numeric. Cluster analysis was an effective means to both distinguish subgroups of CLRD patients with COVID-19 as well as identify patient clusters which were adversely affected by the infection. Age, comorbidity index and race were important factors for cluster separations. Furthermore, diseases of the circulatory system, the nervous system and sense organs, digestive system, genitourinary system, metabolic diseases and immunity disorders were also important criteria in the resulting cluster analyses.