Predicting Long-Term Recovery of Consciousness in Prolonged Disorders of Consciousness Based on Coma Recovery Scale-Revised Subscores: Validation of a Machine Learning-Based Prognostic Index

Abstract

Prognosis of prolonged Disorders of Consciousness (pDoC) is influenced by patients’ clinical diagnosis and Coma Recovery Scale-Revised (CRS-R) total score. We compared the prognostic accuracy of a novel Consciousness Domain Index (CDI) with that of clinical diagnosis and CRS-R total score, for recovery of full consciousness at 6-, 12-, and 24-months post-injury. The CDI was obtained by a combination of the six CRS-R subscales via an unsupervised machine learning technique. We retrospectively analyzed data on 143 patients with pDoC (75 in Minimally Conscious State; 102 males; median age = 53 years; IQR = 35; time post-injury = 1–3 months) due to different etiologies enrolled in an International Brain Injury Association Disorders of Consciousness Special Interest Group (IBIA DoC-SIG) multicenter longitudinal study. Univariate and multivariate analyses were utilized to assess the association between outcomes and the CDI, compared to clinical diagnosis and CRS-R. The CDI, the clinical diagnosis, and the CRS-R total score were significantly associated with a good outcome at 6, 12 and 24 months. The CDI showed the highest univariate prediction accuracy and sensitivity, and regression models including the CDI provided the highest values of explained variance. A combined scoring system of the CRS-R subscales by unsupervised machine learning may improve clinical ability to predict recovery of consciousness in patients with pDoC.

Long‐term clinical evolution of patients with prolonged disorders of consciousness due to severe anoxic brain injury: A meta‐analytic study

“…The selected studies included an overall sample of 2844 patients with disorders of consciousness (eight in coma, 0.3%; 1774 in VS/UWS, 62.4%; 926 in MCS, 32.6%; 136 having emerged from MCS [eMCS], 4.7%) caused by different aetiologies (traumatic brain injury 992, 34.9%; non‐traumatic brain injury 1852, 65.1%), with a mean age of 48 years (not available in two studies); 1825 (64.2%) patients were male (not available in one study). Most studies [3, 5, 13–19, 21, 23, 24, 26, 31, 32, 35, 36, 39, 40] investigated the prognostic role of demographic, clinical and anamnestic factors. Some studies also included neurophysiological [5, 16, 19, 20, 22, 23, 25, 28, 30, 32, 33, 38] or neuroimaging [15, 26, 29, 37–39] investigations.…”

“…Twenty‐three out of the 27 included studies reported all outcomes of interest. Three studies provided data from the same database, but Estraneo et al [5] reported data on patients' improvement at 6 months post‐injury, Estraneo et al [16] reported data on patients' mortality at 24 months post‐injury and Magliacano et al [17] only reported data on recovery of full consciousness at 24 months post‐injury. Moreover, Nekrasova et al [18] only reported data on patients' mortality (see PRISMA flow diagram in Figure 1).…”

“…The details of the included studies are reported in Table S1. Most studies [3,5,14,[16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35] enrolled patients in Europe (23/30, 77%), whereas five studies [15,[36][37][38][39] (16%) included patients from China, and two studies [13,40] (7%) included patients from the Americas.…”

Which information derived from the Coma Recovery Scale-Revised provides the most reliable prediction of clinical diagnosis and recovery of consciousness? A comparative study using machine learning techniques

“…Indices and measures derived from the CRS-R, including the CRS-R total score, the total scores in the subscales of the CRS-R, the clinical diagnosis, the CRS-R Modified Score, 16 the CRS-R Index, 17 the CRS+ 23 and the Consciousness Domain Index, 24 were estimated from the collected data as described in the corresponding manuscripts (Supplementary Digital Material 1: Supplementary Figure 1, 2, Supplementary Table I, II). It is worth to be noted that both CRS-R Modified Score and CRS-R Index calculated in this paper were approximations obtained from the scores on the subscales, instead of evaluating the neurobehavioral signs as required for their derivation.…”

“… 23 More recently, Magliacano and Liuzzi proposed the Consciousness Domain Index, an unsupervised machine learning clustering technique based on information from the CRS-R sub-scales to improve the prediction of recovery of consciousness. 24 …”