Machine Learning Models for Predicting Dysphonia Following Anterior Cervical Discectomy and Fusion - A Swedish Registry Study

Sep 17, 2026

Upload by Asher

Dysphonia is a common complication after anterior cervical discectomy and fusion (ACDF), making early identification of high-risk patients critical. This retrospective Swedish registry study evaluated multiple machine learning approaches to predict persistent dysphonia lasting at least 1 month after surgery. Adults undergoing elective ACDF from 2006 to 2020 were analyzed, excluding patients with missing 1-year dysphonia data.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Abstract

Background

Purpose

Study design

Methods

Results

Conclusions

Zurich Open Repository and Archive University of Zurich University Library Strickhofstrasse 39 CH-8057 Zurich [www.zora.uzh.ch](www.zora.uzh.ch) Year: 2024 Machine Learning Models for Predicting Dysphonia Following Anterior Cervical Discectomy and Fusion : A Swedish Registry Study Buwaider, Ali ; El-Hajj, Victor Gabriel ; MacDowall, Anna ; Gerdhem, Paul ; Staartjes, Victor E ; Edström, Erik ; Elmi-Terander, Adrian DOI: [https://doi.org/10.1016/j.spinee.2024.10.010](https://doi.org/10.1016/j.spinee.2024.10.010) Posted at the Zurich Open Repository and Archive, University of Zurich ZORA URL: [https://doi.org/10.5167/uzh-267253](https://doi.org/10.5167/uzh-267253) Journal Article Accepted Version The following work is licensed under a Creative Commons: Attribution 4.0 International (CC BY 4.0) License. Originally published at: Buwaider, Ali; El-Hajj, Victor Gabriel; MacDowall, Anna; Gerdhem, Paul; Staartjes, Victor E; Edström, Erik; Elmi-Terander, Adrian (2024) . Machine Learning Models for Predicting Dysphonia Following Anterior Cervical Discectomy and Fusion : A Swedish Registry Study. The Spine Journal:Epub ahead of print. DOI: [https://doi.org/10.1016/j.spinee.2024.10.010](https://doi.org/10.1016/j.spinee.2024.10.010) ARTICLE IN PRESS The Spine Journal 000 (2024) 1􀀕10 Clinical Study Machine learning models for predicting dysphonia following anterior cervical discectomy and fusion: a Swedish Registry Study Ali Buwaidera, 1, Victor Gabriel El-Hajja, 1, Anna MacDowall, MD, PhDb,c, Paul Gerdhem, MD, PhDb,c,d, Victor E Staartjes, MD, PhDe, Erik Edstr€om, MD, PhDa,f,g, Adrian Elmi-Terander, MD, PhDa,b,f,g, * a Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden b Department of Surgical Sciences, Uppsala University, Uppsala, Sweden c Department of Orthopaedics and Hand surgery, Uppsala University Hospital, Uppsala, Sweden d Clinical Science, Intervention and Technology (CLINTEC), Karolinska Institutet, Stockholm, Sweden e Machine Intelligence in Clinical Neuroscience (MICN) Laboratory, Department of Neurosurgery, Clinical Neuroscience Center, University Hospital Zurich, University of Zurich, Frauenklinikstrasse 10, 8091, Zurich, Switzerland f Capio Spine Center Stockholm, L€owenstr€omska Hospital, Upplands-V€asby, Sweden g Department of Medical Sciences, €Orebro University, €Orebro, Sweden Received 23 June 2024; revised 27 September 2024; accepted 27 October 2024 Abstract BACKGROUND: Dysphonia is one of the more common complications following anterior cervi cal discectomy and fusion (ACDF) . ACDF is the gold standard for treating degenerative cervical spine disorders, and identifying high-risk patients is therefore crucial. PURPOSE: This study aimed to evaluate different machine learning models to predict persistent dysphonia after ACDF. STUDY DESIGN: A retrospective review of the nationwide Swedish spine registry (Swespine) . PATIENT SAMPLE: All adults in the Swespine registry who underwent elective ACDF between 2006 and 2020 . OUTCOME MEASURES: The primary outcome was self-reported dysphonia lasting at least 1 month after surgery. Predictive performance was assessed using discrimination and calibration metrics. METHODS: Patients with missing dysphonia data at the 1-year follow-up were excluded. Data preprocessing involved one-hot encoding categorical variables, scaling continuous variables, and imputing missing values. Four machine learning models (logistic regression, random forest (RF), gradient boosting, K-nearest neighbor) were employed. The models were trained and tested using an 80:20 data split and 5-fold cross-validation, with performance metrics guiding the selection of the best model for predicting persistent dysphonia. RESULTS: In total, 2,708 were included in the study. Twelve key predictors were identiﬁed. Four machine learning models were tested, with the RF model achieving the best performance (AUC=0.794) . The most signiﬁcant pred

FAQs

What complication does the study aim to predict after ACDF?

The study targets persistent dysphonia after anterior cervical discectomy and fusion (ACDF), lasting at least 1 month after surgery.

How were participants selected for the Swedish registry analysis?

All adults who underwent elective ACDF between 2006 and 2020 in the Swespine registry were included, and patients with missing dysphonia data at the 1-year follow-up were excluded.

Which machine learning model performed best and what does that imply?

Random forest (RF) achieved the best performance (AUC=0.794). It showed high sensitivity and consistent accuracy, though it had low specificity and positive predictive value.

Sep 17, 2026

Upload by Asher