Table 1 Model informationa.

From: Classification of schizophrenia, bipolar disorder and major depressive disorder with comorbid traits and deep learning algorithms

Main Model

Submodel

Target

Traits (# Predictors)

Structure

Baseline(B)

I

SCZ; BIP; MDD

target bestPRS (1)

logistic regression

Baseline(B)

II

SCZ; BIP; MDD

target bestPRS + comorbid bestPRSs (42)

elastic net regression

Baseline(B)

III

SCZ; BIP; MDD

exclusion of target bestPRSs, SCZ (35); BIP (33); MDD (36)

elastic net regression

I

 

SCZ; BIP; MDD

all PRSs, SCZ (237); BIP (237); MDD (237)

deep neural network

II

 

SCZ; BIP; MDD

target bestPRS + comorbid bestPRSs, SCZ (42), BIP (42); MDD (42)

deep neural network

III

 

SCZ; BIP; MDD

exclusion of target PRSs, SCZ (196); BIP (184); MDD (202)

deep neural network

IV

 

SCZ; BIP; MDD

exclusion of target bestPRSs, SCZ (35); BIP (33); MDD (36)

deep neural network

V

I

CTRL, SCZ, BIP and MDD

all PRSs (237)

deep neural network

V

II

CTRL, SCZ, BIP and MDD

target bestPRS + comorbid bestPRSs (42)

deep neural network

V

III

CTRL, SCZ, BIP and MDD

exclusion of target PRSs (166)

deep neural network

V

IV

CTRL, SCZ, BIP and MDD

exclusion of target bestPRSs (30)

deep neural network

  1. aModel nomenclature: Main_model.submodel.target_disorder. If a main model does not have a submodel, then main_model.target_disorder. For multiclass classification models (main model V), the number of classes would be used as the target. For example, B.I.SCZ, I.SCZ and V.I.4C.
Back to article page