Skip to main content

Table 2 A comparison between previous literature and our proposed models (bolded values) with regard to the classification of a minimum of four in-bed postures, containing at least supine, prone, right, and left

From: A novel in-bed body posture monitoring for decubitus ulcer prevention using body pressure distribution mapping

Year & Refs.

# Subs

# Postures

Algorithm

Cross-Validation Method

Performance

2019 [12]

12

4

FFAN\(^1\)

LOSO

ACC\(^2\): 97.9%

2020 [13]

7

4

kNN\(^3\)

LOSO

ACC: 79.02%

2021 [11]

1

6

Inception-v3

Holdout

ACC: 90.5%

2022 [14]

6

4

SRC\(^4\)

6-Fold & LOSO

ER\(^5\): 0.09

Proposed Models

10

6

I3D DNN \(^6\)

LOSO

ACC: 82.22%

    

LOEO

ACC: 77.78%

   

ResNet-18

LOSO

ACC: 92.07%

    

LOEO

ACC: 85.37%

  1. \(^1\)FFAN: Feed-Forward Artificial Neural Network, \(^2\)ACC: Accuracy, \(^3\)kNN: k-Nearest Neighbor, \(^4\)SRC: Sparse Representation Classification, \(^5\)ER: Error Rate, \(^6\)DNN: Deep Neural Network