Medical decision support using machine learning for early detection of late-onset neonatal sepsis

Subramani Mani, Asli Ozdas, Constantin Aliferis, Huseyin Atakan Varol, Qingxia Chen, Randy Carnevale, Yukun Chen, Joann Romano-Keeler, Hui Nian, Jörn Hendrik Weitkamp

    Research output: Contribution to journalArticle

    38 Citations (Scopus)

    Abstract

    Objective: The objective was to develop non-invasive predictive models for late-onset neonatal sepsis from offthe-shelf medical data and electronic medical records (EMR). Design: The data used in this study are from 299 infants admitted to the neonatal intensive care unit in the Monroe Carell Jr. Children's Hospital at Vanderbilt and evaluated for late-onset sepsis. Gold standard diagnostic labels (sepsis negative, culture positive sepsis, culture negative/clinical sepsis) were assigned based on all the laboratory, clinical and microbiology data available in EMR. Only data that were available up to 12 h after phlebotomy for blood culture testing were used to build predictive models using machine learning (ML) algorithms. Measurement: We compared sensitivity, specificity, positive predictive value and negative predictive value of sepsis treatment of physicians with the predictions of models generated by ML algorithms. Results: The treatment sensitivity of all the nine ML algorithms and specificity of eight out of the nine ML algorithms tested exceeded that of the physician when culture-negative sepsis was included. When culturenegative sepsis was excluded both sensitivity and specificity exceeded that of the physician for all the ML algorithms. The top three predictive variables were the hematocrit or packed cell volume, chorioamnionitis and respiratory rate. Conclusions: Predictive models developed from off-theshelf and EMR data using ML algorithms exceeded the treatment sensitivity and treatment specificity of clinicians. A prospective study is warranted to assess the clinical utility of the ML algorithms in improving the accuracy of antibiotic use in the management of neonatal sepsis.

    Original languageEnglish
    Pages (from-to)326-336
    Number of pages11
    JournalJournal of the American Medical Informatics Association
    Volume21
    Issue number2
    DOIs
    Publication statusPublished - 2014

    Fingerprint

    Sepsis
    Electronic Health Records
    Physicians
    Sensitivity and Specificity
    Medical Electronics
    Chorioamnionitis
    Phlebotomy
    Neonatal Intensive Care Units
    Therapeutics
    Respiratory Rate
    Microbiology
    Neonatal Sepsis
    Machine Learning
    Cell Size
    Hematocrit
    Prospective Studies
    Anti-Bacterial Agents

    ASJC Scopus subject areas

    • Health Informatics

    Cite this

    Medical decision support using machine learning for early detection of late-onset neonatal sepsis. / Mani, Subramani; Ozdas, Asli; Aliferis, Constantin; Varol, Huseyin Atakan; Chen, Qingxia; Carnevale, Randy; Chen, Yukun; Romano-Keeler, Joann; Nian, Hui; Weitkamp, Jörn Hendrik.

    In: Journal of the American Medical Informatics Association, Vol. 21, No. 2, 2014, p. 326-336.

    Research output: Contribution to journalArticle

    Mani, S, Ozdas, A, Aliferis, C, Varol, HA, Chen, Q, Carnevale, R, Chen, Y, Romano-Keeler, J, Nian, H & Weitkamp, JH 2014, 'Medical decision support using machine learning for early detection of late-onset neonatal sepsis', Journal of the American Medical Informatics Association, vol. 21, no. 2, pp. 326-336. https://doi.org/10.1136/amiajnl-2013-001854
    Mani, Subramani ; Ozdas, Asli ; Aliferis, Constantin ; Varol, Huseyin Atakan ; Chen, Qingxia ; Carnevale, Randy ; Chen, Yukun ; Romano-Keeler, Joann ; Nian, Hui ; Weitkamp, Jörn Hendrik. / Medical decision support using machine learning for early detection of late-onset neonatal sepsis. In: Journal of the American Medical Informatics Association. 2014 ; Vol. 21, No. 2. pp. 326-336.
    @article{64672b8ef32744f28badcb4fcb9c3fea,
    title = "Medical decision support using machine learning for early detection of late-onset neonatal sepsis",
    abstract = "Objective: The objective was to develop non-invasive predictive models for late-onset neonatal sepsis from offthe-shelf medical data and electronic medical records (EMR). Design: The data used in this study are from 299 infants admitted to the neonatal intensive care unit in the Monroe Carell Jr. Children's Hospital at Vanderbilt and evaluated for late-onset sepsis. Gold standard diagnostic labels (sepsis negative, culture positive sepsis, culture negative/clinical sepsis) were assigned based on all the laboratory, clinical and microbiology data available in EMR. Only data that were available up to 12 h after phlebotomy for blood culture testing were used to build predictive models using machine learning (ML) algorithms. Measurement: We compared sensitivity, specificity, positive predictive value and negative predictive value of sepsis treatment of physicians with the predictions of models generated by ML algorithms. Results: The treatment sensitivity of all the nine ML algorithms and specificity of eight out of the nine ML algorithms tested exceeded that of the physician when culture-negative sepsis was included. When culturenegative sepsis was excluded both sensitivity and specificity exceeded that of the physician for all the ML algorithms. The top three predictive variables were the hematocrit or packed cell volume, chorioamnionitis and respiratory rate. Conclusions: Predictive models developed from off-theshelf and EMR data using ML algorithms exceeded the treatment sensitivity and treatment specificity of clinicians. A prospective study is warranted to assess the clinical utility of the ML algorithms in improving the accuracy of antibiotic use in the management of neonatal sepsis.",
    author = "Subramani Mani and Asli Ozdas and Constantin Aliferis and Varol, {Huseyin Atakan} and Qingxia Chen and Randy Carnevale and Yukun Chen and Joann Romano-Keeler and Hui Nian and Weitkamp, {J{\"o}rn Hendrik}",
    year = "2014",
    doi = "10.1136/amiajnl-2013-001854",
    language = "English",
    volume = "21",
    pages = "326--336",
    journal = "Journal of the American Medical Informatics Association : JAMIA",
    issn = "1067-5027",
    publisher = "Oxford University Press",
    number = "2",

    }

    TY - JOUR

    T1 - Medical decision support using machine learning for early detection of late-onset neonatal sepsis

    AU - Mani, Subramani

    AU - Ozdas, Asli

    AU - Aliferis, Constantin

    AU - Varol, Huseyin Atakan

    AU - Chen, Qingxia

    AU - Carnevale, Randy

    AU - Chen, Yukun

    AU - Romano-Keeler, Joann

    AU - Nian, Hui

    AU - Weitkamp, Jörn Hendrik

    PY - 2014

    Y1 - 2014

    N2 - Objective: The objective was to develop non-invasive predictive models for late-onset neonatal sepsis from offthe-shelf medical data and electronic medical records (EMR). Design: The data used in this study are from 299 infants admitted to the neonatal intensive care unit in the Monroe Carell Jr. Children's Hospital at Vanderbilt and evaluated for late-onset sepsis. Gold standard diagnostic labels (sepsis negative, culture positive sepsis, culture negative/clinical sepsis) were assigned based on all the laboratory, clinical and microbiology data available in EMR. Only data that were available up to 12 h after phlebotomy for blood culture testing were used to build predictive models using machine learning (ML) algorithms. Measurement: We compared sensitivity, specificity, positive predictive value and negative predictive value of sepsis treatment of physicians with the predictions of models generated by ML algorithms. Results: The treatment sensitivity of all the nine ML algorithms and specificity of eight out of the nine ML algorithms tested exceeded that of the physician when culture-negative sepsis was included. When culturenegative sepsis was excluded both sensitivity and specificity exceeded that of the physician for all the ML algorithms. The top three predictive variables were the hematocrit or packed cell volume, chorioamnionitis and respiratory rate. Conclusions: Predictive models developed from off-theshelf and EMR data using ML algorithms exceeded the treatment sensitivity and treatment specificity of clinicians. A prospective study is warranted to assess the clinical utility of the ML algorithms in improving the accuracy of antibiotic use in the management of neonatal sepsis.

    AB - Objective: The objective was to develop non-invasive predictive models for late-onset neonatal sepsis from offthe-shelf medical data and electronic medical records (EMR). Design: The data used in this study are from 299 infants admitted to the neonatal intensive care unit in the Monroe Carell Jr. Children's Hospital at Vanderbilt and evaluated for late-onset sepsis. Gold standard diagnostic labels (sepsis negative, culture positive sepsis, culture negative/clinical sepsis) were assigned based on all the laboratory, clinical and microbiology data available in EMR. Only data that were available up to 12 h after phlebotomy for blood culture testing were used to build predictive models using machine learning (ML) algorithms. Measurement: We compared sensitivity, specificity, positive predictive value and negative predictive value of sepsis treatment of physicians with the predictions of models generated by ML algorithms. Results: The treatment sensitivity of all the nine ML algorithms and specificity of eight out of the nine ML algorithms tested exceeded that of the physician when culture-negative sepsis was included. When culturenegative sepsis was excluded both sensitivity and specificity exceeded that of the physician for all the ML algorithms. The top three predictive variables were the hematocrit or packed cell volume, chorioamnionitis and respiratory rate. Conclusions: Predictive models developed from off-theshelf and EMR data using ML algorithms exceeded the treatment sensitivity and treatment specificity of clinicians. A prospective study is warranted to assess the clinical utility of the ML algorithms in improving the accuracy of antibiotic use in the management of neonatal sepsis.

    UR - http://www.scopus.com/inward/record.url?scp=84894094760&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84894094760&partnerID=8YFLogxK

    U2 - 10.1136/amiajnl-2013-001854

    DO - 10.1136/amiajnl-2013-001854

    M3 - Article

    VL - 21

    SP - 326

    EP - 336

    JO - Journal of the American Medical Informatics Association : JAMIA

    JF - Journal of the American Medical Informatics Association : JAMIA

    SN - 1067-5027

    IS - 2

    ER -