Back

From Tabulated Data to Knowledge Graph: A Novel Way of Improving the Performance of the Classification Models in the Healthcare Data

Zaki, N.; Mohamed, E. A.; Habuza, T.

2021-06-12 health informatics
10.1101/2021.06.09.21258123 medRxiv
Show abstract

In sectors like healthcare, having classification models that are both reliable and accurate is vital. Regrettably, contemporary classification techniques employing machine learning disregard the correlations between instances within data. This research, to rectify this, introduces a basic but effective technique for converting tabulated data into data graphs, incorporating structural correlations. Graphs have a unique capacity to capture structural correlations between data, allowing us to gain a deeper insight in comparison to carrying out isolated data analysis. The suggested technique underwent testing once the integration of graph data structure-related elements had been carried out and returned superior results to testing solely employing original features. The suggested technique achieved validity by returning significantly improved levels of accuracy. DataThe extracted graph topological features datasets are available from:

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.