Outcome of Severe Vaso-Occlusive Crisis in Sickle Cell Disease Adults Admitted to Referral Centers in Africa and Europe. Introduction of Machine Learning Methods to Improve the Presev Score

Blood ◽  
2020 ◽  
Vol 136 (Supplement 1) ◽  
pp. 14-15
Author(s):  
Yamna Ouchtar ◽  
Christian Kassasseya ◽  
Kene Sekou ◽  
Anne-Laure Pham Hung D'Alexandry D'Orengiani ◽  
Mehdi Kellaf ◽  
...  
Keyword(s):  
Machine Learning ◽  
Sickle Cell Disease ◽  
Decision Tree ◽  
Predictive Value ◽  
Research Funding ◽  
Low Risk ◽  
Cell Disease ◽  
Risk Patients ◽  
Decision Tree Method ◽  
Tree Method

Introduction: Sickle Cell Disease (SCD) is one of the most common genetic disease worldwide. The Acute Chest Syndrome (ACS) is a leading cause of death for SCD patients. The PRESEV1 study was set to produce a predictive score to assess the risk of an ACS development (Bartolucci et al., 2016). PRESEV2 was an international, multicenter prospective confirmatory study to validate the PRESEV score. This study aims at improving these predictions with the addition of a machine learning (ML) method. Patients and methods: Included patients follow PRESEV1 and PRESEV2 studies 'rules. The dataset thus contains 97 patients who developed an ACS episode (18.3%) against 434 patients who did not (81.7%). To compute the PRESEV score, we firstly used the method developed previously with the following variables as input: leukocytes, reticulocytes, hemoglobin levels and cervical spine pain. This method is based on a decision tree with fixed rules and is referred to as the decision tree method throughout this abstract. Secondly we used a ML method using a combined sampling method named SMOTEENN to balance the data and a C-Support Vector Classification (SVC) with fixed parameters to predict the score. This method produces a probability, with a threshold of 0.2, under which the patient is predicted to declare an ACS. We considered the dataset composed of PRESEV1 dataset and 80 percent of PRESEV2 with a randomly choice. The test dataset is thus composed of the remaining 20 percent of PRESEV2. This technique of random choice allowed us to use a 50-cross-validation and compute with Python an average score and a standard deviation (std). In order to allow comparison of the developed score with or without the addition of the ML method, rates were calculated by adding the weight of ACS representation in the dataset. Results: Among all parameters analyzed, the SVC method considered the following variables for calculation of the score: leukocytes, LDH, urea, reticulocytes and hemoglobin levels. A hundred and two adult patients with a severe VOC requiring hospitalization were included. Out of this pool of patients, 26 (25.5%) were predicted with a low risk of developing an ACS episode (SVC method). Sensibility and specificity were of 94.7% and 26.8%, respectfully. The negative predictive value (NPV) was of 95.8% and the positive predictive value (PPV) of 22.4%. Results are resumed in table 1. When compared to the PRESEV score (decision tree method), 44 patients out of 372 were identified with a low risk score (11.8%), Discussion and Conclusion: While the addition of a ML method did not allow the improvement of the sensibility or the NPV of the PRESEV score, it improved both the specificity and the PPV. The addition of artificial intelligence thus provides a better prediction with a higher percentage of "low-risk" patients. As highlighted in the international PRESEV study, this score could represent a useful tool for physicians in hospital settings, with limited beds. While the PRESEV score could allow a better management of "low risk" patients on one side, the identification of "high-risk" patients could also represent a serious advantage to physicians, as it could improve the feasibility of clinical trials for the prevention of this lethal complication in SCD patients. Disclosures Bartolucci: Innovhem: Other; Novartis: Research Funding; Roche: Consultancy; Bluebird: Consultancy; Emmaus: Consultancy; Bluebird: Research Funding; Addmedica: Research Funding; AGIOS: Consultancy; Fabre Foundation: Research Funding; Novartis: Consultancy; ADDMEDICA: Consultancy; HEMANEXT: Consultancy; GBT: Consultancy.

scholarly journals Accuracy of Administrative Coding for Sickle Cell Disease

Blood ◽  
2015 ◽  
Vol 126 (23) ◽  
pp. 3260-3260
Author(s):  
Michelle Ting ◽  
Shyamli Sinha ◽  
Timothy L. McCavit
Keyword(s):  
Sickle Cell Disease ◽  
Likelihood Ratio ◽  
Acute Care ◽  
Administrative Data ◽  
Predictive Value ◽  
Research Funding ◽  
Negative Likelihood Ratio ◽  
Senior Author ◽  
Cell Disease ◽  
Inpatient Hospitalization

Abstract Background Administrative data are increasingly used in sickle cell disease (SCD) research to study large numbers of patients at low cost. However, the validity of research with these data depends on the accuracy of administrative coding, which has been understudied in SCD. In particular, the validity of ICD-9CM coding for SCD's clinical hallmark, vaso-occlusive crisis (VOC), has not been reported. Therefore, we aim to describe the accuracy of ICD-9CM coding for VOC, acute chest syndrome (ACS), and acute splenic sequestration crisis (ASSC) in addition to SCD genotypes. Methods Administrative coding for all acute care visits (emergency department [ED], inpatient observation, and inpatient hospitalization) in SCD patients was captured for the 2013 calendar year at Children's Medical Center Dallas (CMCD) by query of administrative records. SCD visits were identified using the ICD-9CM codes 282.4x and 282.6x in the primary or any of 14 secondary code positions. From the administrative data, VOC was defined by the use of the "in crisis" codes (282.42, 282.62, 282.64, or 282.69). ACS was defined by 517.3 and ASSC by 289.52. Genotypes were defined as HbSS - 282.61 or 282.62; HbSC - 282.63 or 282.64; and HbS-BetaThal - 282.41 or 282.42. For the chart review, all visits were independently evaluated by two reviewers with the senior author settling disagreements. Previously-published consensus definitions were used for the complications of interest: VOC was defined as new onset of pain lasting at last 4 hours for which there is no other explanation; ACS was defined by new pulmonary infiltrate on chest X-ray with associated fever or respiratory symptoms; and ASSC was defined as splenic enlargement leading to anemia (hemoglobin >2g/dL under baseline), typically with thrombocytopenia. The primary reviewers and senior author evaluated 10 cases for the purpose of consensus-building before beginning data abstraction. Data capture included demographics and cause of the visit. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (LR+), and negative likelihood ratio (LR-) of ICD-9CM coding for VOC, ACS, ASSC, and the SCD genotype were calculated. The LR+ and LR- were prevalence-weighted. Results From 448 patients, 1107 acute care visits were identified at CMCD in 2013. Patients had a median age of 8.2 years (range 0.1 - 33.9 years). Females accounted for 53% of visits. Inpatient hospitalization accounted for 374 visits (34%) whereas ED-only visits accounted for 681(61%) and inpatient observation 52 (5%). The prevalence of SCD genotypes included 69% HbSS, 19% HbSC, and 7% HbS-BetaThal. VOC, ACS, and ASSC were the "true cause" for 36% (n=402), 6% (n=68), and 1% (n=13) of acute care visits, respectively. The results of the primary analyses are displayed in the table. Agreement between reviewers for VOC and ACS was 91% (kappa = 0.80) and 99% (kappa = 0.92), respectively. When VOC was falsely coded (n=176), the most common true diagnoses were ACS (26%), fever (24%), chronic pain exacerbation (14%), and abdominal pain (10%). When ACS was falsely coded (n=21), VOC was the most common true diagnosis (71%). Of 25 (2.2%) patients incorrectly coded for SCD, 14 had sickle trait. Conclusions ICD-9CM coding for VOC demonstrated poor accuracy; however, coding for ACS and ASSC was remarkably sensitive and specific. Genotype coding lacked sensitivity with otherwise variable results. Unfortunately, coding for SCD in ICD-10 differs minimally from ICD-9CM. Therefore, these data provide an impetus to restructure ICD coding for SCD. Table 1. Sensitivity (95% CI) Specificity (95% CI) PPV(95% CI) NPV(95% CI) Postive Likelihood Ratio(95% CI) Negative Likelihood Ratio(95% CI) VOC 91.8(88.6-94.2) 75.0(71.6-78.2) 67.7(63.6-71.6) 94.1(91.8-95.9) 2.1(1.8-2.4) 0.06(0.04-0.08) ACS 92.6(83.0-97.3) 98.0(96.9-98.7) 75.0(64.1-83.5) 99.5(98.8-99.8) 3.0(2.0-4.4) 0.004(0.002-0.01) ASSC 92.3(62.1-99.6) 99.8(99.3-100.0) 85.7(56.2-97.5) 99.9(99.4-100.0) 6.0(1.6-22.0) 0.0009(0.0001-0.006) HbSS genotype 74.2(70.9-77.2) 67.7(62.5-72.6) 83.8(80.7-86.4) 53.8(49.0-58.6) 5.1 (4.2-6.1) 0.9(0.8-1.0) HbSC genotype 31.0(24.9-37.7) 99.6(98.8-99.9) 94.3(85.3-98.2) 85.8(83.5-87.9) 16.5(6.4-42.8) 0.16(0.14-0.19) HbS-BetaThalgenotype 57.1(45.4-68.2) 99.6(98.9-99.9) 91.7(79.1-97.3) 96.9(95.6-97.8) 11(4.3-28.2) 0.03(0.02-0.05) Disclosures McCavit: Pfizer: Research Funding; Novartis: Speakers Bureau; Novartis: Speakers Bureau; Gensavis LLC: Research Funding; Pfizer: Research Funding.

scholarly journals Evaluating Age As an Independent Risk Factor for Serious Bacterial Illness in Febrile Children with Sickle Cell Disease

Blood ◽  
2021 ◽  
Vol 138 (Supplement 1) ◽  
pp. 767-767
Author(s):  
Leya Y Schwartz ◽  
Grace Ye ◽  
Daniel M Fein ◽  
Kerry A Morrone
Keyword(s):  
Risk Factor ◽  
Sickle Cell Disease ◽  
Predictive Value ◽  
Independent Risk Factor ◽  
Screening Tool ◽  
Low Risk ◽  
Cell Disease ◽  
Pediatric Hematology ◽  
Test Characteristics ◽  
Sensitivity Specificity

Abstract Children with sickle cell disease (SCD) have increased morbidity and mortality secondary to serious bacterial infections (SBI) by encapsulated organisms. These include bacteremia, acute chest syndrome (ACS), urinary tract infection, meningitis, osteomyelitis, septic arthritis, and cholangitis. For febrile patients with SCD, screening tools have been used in pediatric emergency departments (ED) across the United States. One such screening tool is a clinical pathway utilized at the Children's Hospital of Philadelphia, where low-risk criteria is used to determine risk for SBI and management thereafter (Ellison et al. Journal of Pediatric Hematology/Oncology 2018). Our institution defines low-risk criteria as lacking all of the following; ill-appearance, hypotension, a temperature ≥40 oC, history of Streptococcus pneumoniae bacteremia, diagnosis of ACS, clinical or laboratory concern for splenic sequestration (Hgb decrease by >2g/dL from baseline with an enlarging spleen and/or thrombocytopenia with platelets <100k/uL), white blood cell count >5k/uL and <30k/uL, Hgb <5g/dL, non-adherence with penicillin prophylaxis, cephalosporin allergy, incomplete immunizations, or concerns regarding appropriate follow up. Our institution routinely admits febrile patients, <24 months of age with SCD, irrespective of any other factor. Given the low rate of SBI among febrile children with SCD at low-risk, and the risks associated with hospital admissions, there is likely an opportunity to safely treat younger children similarly to their older counterparts. The primary objective of our study is to compare test characteristics, specifically the negative predictive value (NPV), of our screening tool (not including age) on predicting risk of SBI in febrile children with SCD ages 6-24 months and those older than 24 months. This 10-year retrospective cohort study included patients with SCD <21 years old who had fever (≥38 oC, during their visit or on history prior to arrival) and were seen in the ED or Pediatric Hematology outpatient clinic at our institution. Test characteristics (sensitivity, specificity, positive predictive value (PPV) and NPV) of our screening tool were calculated for both age groups. Pearson's chi square was utilized to compare the test characteristics between the two groups. A total of 1226 encounters were analyzed, 320 (26%) of those were patients 6-24 months of age and 877 (72%) of those were patients >24 months. There were 35 (11%) patients in the 6-24 month cohort that had an SBI and 201 (23%) patients in the >24 month cohort that had an SBI. The sensitivity, specificity, PPV and NPV for both age cohorts are represented in Table 1. There was no statistically significant difference between the NPV of those 6-24 months of age 98% (95% CI 95%-99%) and the NPV of those >24 months of age 97% (95% CI 95%-98%) with a p-value of 0.63. In conclusion, the screening tool performed well in identifying children with SCD and fever who are at low-risk of SBI in both age ranges. There was no difference in its performance in the younger group suggesting that a cut-off age of 2 years is not an independent risk factor for SBI and that children 6-24 months should not be admitted solely because of their age. Figure 1 Figure 1. Disclosures No relevant conflicts of interest to declare.

scholarly journals Outcomes of Fever in Infants and Young Children with SCD Presenting to the Emergency Room

Blood ◽  
2016 ◽  
Vol 128 (22) ◽  
pp. 3591-3591
Author(s):  
Krishnaveni Sirigaddi ◽  
Inmaculada Aban ◽  
Christina J. Bemrich-Stolz ◽  
Thomas H. Howard ◽  
Lee Hilliard ◽  
...  
Keyword(s):  
High Risk ◽  
Sickle Cell Disease ◽  
Young Children ◽  
Emergency Room ◽  
Sensitivity And Specificity ◽  
Predictive Models ◽  
Research Funding ◽  
Distributed Data ◽  
Cell Disease ◽  
Risk Patients

Abstract The introduction of antibiotic prophylaxis and vaccinations has reduced the incidence of bacteremia and sepsis in pediatric patients with sickle cell disease (SCD). Due to concern of mortality from sepsis, SCD patients with fever require admission for IV antibiotics until bacteremia is ruled out. The 2014 NIH Evidence-Based Management of Sickle Cell Disease guidelines recommends hospitalization for patients with temperature ≥ 39.5 C and who appear ill. Prior pediatric research has defined "high risk patients" as those with temperature ≥ 40 C, WBC >30,000 or <5,000/mcl, appear ill, or have pulmonary infiltrates on chest x-ray (CXR). (Wilimas NEJM 1993) We reviewed all Emergency Room (ER) visits for febrile patients <6 years of age with HbSS or HbSB0 thalassemia to evaluate these predictive models for bacteremia. Methods: In a 16 year IRB approved cohort, we identified 609 ER visits to Children's of Alabama for fever among 169 children < 6 years old with HbSS or SB0 thalassemia. All patients receive standard of care penicillin prophylaxis and vaccination. We reviewed every blood culture obtained during their ER visit and admission to determine the incidence of bacteremia including pneumococcal bacteremia. We recorded vital signs, blood counts, and CXR findings during the ER visit. We compared differences in these variables among patients with and without bacteremia. We created categorical variables (yes/no) to evaluate NIH fever guidelines (Temp >39.5C and ill appearance) and "high risk patient" recommendations (Temp >40C, ill appearance, abnormal white blood cell count (>30,000 or <5,000//mcl), or pulmonary infiltrates on CXR). Descriptive statistics, t-test for normally distributed data and Wilcoxon for non-normally distributed data, and Fisher's exact test were performed in JMP12. Sensitivity and specificity were calculated from 2x2 tables. To examine the predictive models for bacteremia, we utilized multiple logistic regression to develop the receiving operating characteristics curves and area under the curve (AUC). Results: Among the 169 patients (0-5.99yrs) with HbSS or SB0 thalassemia that were evaluated in the Children's of Alabama ER for fever, 95 (56%) were female. Five hundred and twelve (84%) admissions were identified among the 609 ER visits including all patients with bacteremia. Fourteen patients (2.3%) evaluated in the ER for fever were subsequently diagnosed with bacteremia including 9 (1.5%) positive for pneumococcus. The incidence of bacteremia among young patients presenting to the ER for fever was 1.4 events per 100 patient years and the rate of pneumococcus was 0.9 events per 100 patient years. Patients with bacteremia had higher WBC (27.0 ±7.8 vs 17.2 ±8.5, p<0.0001) than patients without bacteremia. No statistical differences were noted for patients with and without bacteremia for temperature (p=0.06) or heart rate (p=0.3). The sensitivity and specificity of individual variables for bacteremia were: Temp ≥ 39.5 (Sen: 57%, Specificity: 65%), Temp ≥ 40 (sensitivity: 29%, specificity 90%), Ill appearing (sensitivity: 43%, specificity: 86%), abnormal WBC (sensitivity: 36%, specificity 91%) abnormal CXR (appearing (sensitivity: 57%, specificity: 72%). To evaluate models for bacteremia, the AUC for NIH admission guidelines and "high risk patients" were 0.68 and 0.80 respectively. Conclusion: While the incidence of bacteremia is low, young children with SCD are frequently admitted for IV antibiotics until bacteremia is ruled out. Our data suggests using the "high risk model" for admission criteria in febrile children with SCD. Developing models that can accurately predict bacteremia are limited due to the low incidence of bacteremia. Disclosures Lebensburger: NHLBI: Research Funding; American Society of Hematology, Scholar Award: Research Funding.

Rancang Bangun Aplikasi Sistem Pakar untuk Menangani Masalah Kecantikan pada Wajah Menggunakan Metode Decision Tree

2014 ◽  
Vol 6 (1) ◽  
pp. 9-14
Author(s):  
Stefanie Sirapanji ◽  
Seng Hansun
Keyword(s):  
Decision Tree ◽  
Uneven Distribution ◽  
Dry Skin ◽  
Average Accuracy ◽  
Accuracy Level ◽  
Index Terms ◽  
Decision Tree Method ◽  
Tree Method

Beauty is a precious asset for everyone. Everyone wants to have a healthy face. Unfortunately, there are always those problems that pops out on its own. For example, acnes, freckles, wrinkles, dull, oily and dry skin. Therefore, nowadays, there are a lot of beauty clinics available to help those who wants to solve their beauty troubles. But, not everyone can enjoy the facilities of those beauty clinics, for example those in the suburbs. The uneven distribution of doctors and the expensive cost of treatments are some of the reasons. In this research, the system that could help the patients to find the solution of their beauty problems is built. The decision tree method is used to take decision based on the shown schematic. Based on the system’s experiment, the average accuracy level hits 100%. Index Terms–Acnes, Decision Tree, Dry Skin, Dull, Facial Problems, Freckles, Wrinkles, Oily Skin, Eexpert System.

scholarly journals A Novel Epigenetic Machine Learning Model to Define Risk of Progression for Hepatocellular Carcinoma Patients

2021 ◽  
Vol 22 (3) ◽  
pp. 1075
Author(s):  
Luca Bedon ◽  
Michele Dal Bo ◽  
Monica Mossenta ◽  
Davide Busato ◽  
Giuseppe Toffoli ◽  
...  
Keyword(s):  
Machine Learning ◽  
Hepatocellular Carcinoma ◽  
High Risk ◽  
Progression Free Survival ◽  
Low Risk ◽  
Functional Enrichment ◽  
Free Survival ◽  
High Risk Patients ◽  
Risk Of Progression ◽  
Risk Patients

Although extensive advancements have been made in treatment against hepatocellular carcinoma (HCC), the prognosis of HCC patients remains unsatisfied. It is now clearly established that extensive epigenetic changes act as a driver in human tumors. This study exploits HCC epigenetic deregulation to define a novel prognostic model for monitoring the progression of HCC. We analyzed the genome-wide DNA methylation profile of 374 primary tumor specimens using the Illumina 450 K array data from The Cancer Genome Atlas. We initially used a novel combination of Machine Learning algorithms (Recursive Features Selection, Boruta) to capture early tumor progression features. The subsets of probes obtained were used to train and validate Random Forest models to predict a Progression Free Survival greater or less than 6 months. The model based on 34 epigenetic probes showed the best performance, scoring 0.80 accuracy and 0.51 Matthews Correlation Coefficient on testset. Then, we generated and validated a progression signature based on 4 methylation probes capable of stratifying HCC patients at high and low risk of progression. Survival analysis showed that high risk patients are characterized by a poorer progression free survival compared to low risk patients. Moreover, decision curve analysis confirmed the strength of this predictive tool over conventional clinical parameters. Functional enrichment analysis highlighted that high risk patients differentiated themselves by the upregulation of proliferative pathways. Ultimately, we propose the oncogenic MCM2 gene as a methylation-driven gene of which the representative epigenetic markers could serve both as predictive and prognostic markers. Briefly, our work provides several potential HCC progression epigenetic biomarkers as well as a new signature that may enhance patients surveillance and advances in personalized treatment.

Marketing Model Investigation of Cable Television Service Based on BP Neural Network and Decision Tree

2013 ◽  
Vol 774-776 ◽  
pp. 1757-1761
Author(s):  
Bing Xiang Liu ◽  
Xu Dong Wu ◽  
Ying Xi Li ◽  
Xie Wei Wang
Keyword(s):  
Neural Network ◽  
Decision Tree ◽  
Bp Neural Network ◽  
Cable Television ◽  
Model Building ◽  
Television System ◽  
Marketing Model ◽  
Television Service ◽  
Decision Tree Method ◽  
Tree Method

This paper takes more than four hundred records of some cable television system for example, makes data mining according to users data record, uses BP neural network and decision tree method respectively to have model building and finds the best model fits for users to order press service. The results of the experiment validate the methods feasibility and validity.

Sign in / Sign up

Export Citation Format

Share Document