For this dataset, our algorithm achieves the prediction accuracy is 97.8% (1265 structures were correctly perceived).
Moreover, they used 190 mixed Markov-chain models to achieve the prediction accuracy of 64%%, whilst one Markov-chain model with one transition matrix for the 691 participants was able to achieve a prediction accuracy of 45.6 %.
And, when the number of users is 90, the accuracy of the recommendation algorithm based on the ant colony algorithm has been unable to achieve the prediction of the recommended content.
Minimally, this would achieve the prediction performance of the convex hull of the ROC curves shown in Figure 4. Still, such a consensus would have an AUC value of less than 0.70, which is conventionally accepted as a usable prediction performance of sequence based prediction methods.
So why is it so difficult to achieve the prediction?
The objective of our study was to achieve the prediction accuracy within 1 mm for prediction horizon tpred = 1 second by using NN approaches and to study capabilities of the simplest yet powerful NN models.
We believe that our approach is a significant success, considering that we only used five gene markers in a panel to achieve the prediction performance (AUC = 0.7879, precision = 72.41%, accuracy = 68.75%, sensitivity = 63.64%, and specificity = 74.19%).
We believe our approach is a significant success, considering that we only used fifteen gene markers in a panel to achieve the prediction performance (AUC = 0.8, Precision = 74.29%, Accuracy = 73.85%, Sensitivity = 76.47%, Specificity = 70.97%).
A 150-min prediction horizon used on ten patients and the system achieved the best prediction results of 73% sensitivity and 67% specificity.
