Automated Surgical Term Clustering: A Text Mining Approach for Unstructured Textual Surgery Descriptions

Khaleghi, Tannaz; Murat, Alper; Arslanturk, Suzan; Davies, Eric

doi:10.1109/jbhi.2019.2956973

Cited by 12 publications

(13 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This distance method calculation is represented as L in the algorithm. The T parameter is a threshold distance measure defined specifically for each specialty based on the level of dissimilarity in description words [9].…”

Section: Class Weight Recalculation Methodsmentioning

confidence: 99%

“…Unstructured data, such as textual features of procedure description and notes, provide additional information while structured data alone is not sufficient. We extract the TFIDF scores of the terms (i.e., text features) by deploying our text mining and text feature extraction method [9]. The consistency of the data reflects how close the prediction will be to the ground truth labels obtained from post-surgery medical coding done by medical coding professionals.…”

Section: Input Datamentioning

confidence: 99%

“…There-fore, any prediction attempt directly from unprocessed text simply results in poor estimations with high prediction error. In our prior study, we have identified reduced and structured feature sets by utilizing an un-supervised text mining approach from the free-text descriptions [9,5].…”

Section: Introductionmentioning

confidence: 99%

“…While this study benefits from using the extracted features from free-text in our previous research [9], it integrates them with other continuous and categorical features to classify the surgeries by the primary CPT codes. Incorporating textual information available pre-surgery is shown to drastically improve CPT prediction [9]. In this study, given the multi-attribute feature space, the CPT codes are predicted by Random Forest (RF) classification model tuned with optimal parameters.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

A Tree Based Approach for Multi-class Classification of Surgical Procedures Using Structured and Unstructured Data

Khaleghi

Murat

Arslanturk

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Background: In surgical department, CPT code assignment has been a complicated manual human effort, that entails significant related knowledge and experience. While there are several studies using CPTs to make predictions in surgical services, literature on predicting CPTs in surgical and other services using text features is very sparse. This study improves the prediction of CPTs by the means of informative features and novel re-prioritization algorithm. Methods: The input data used in this study is composed of both structured and unstructured data. The ground truth labels (CPTs) are obtained from medical coding databases using relative value units which indicates the major operational procedures in each surgery case. In the modeling process, we first utilize Random Forest multi-class classification model to predict the CPT codes. Second, we extract the key information such as label probabilities, feature importance measures, and medical term frequency. Then, the indicated factors are used in a novel algorithm to rearrange the alternative CPT codes in the list of potential candidates based on the calculated weights. Results: To evaluate the performance of both phases, prediction and complementary improvement, we report the accuracy scores of multi-class CPT prediction tasks for datasets of 5 key surgery case specialities. The Random Forest model performs the classification task with 74% to 76% when predicting the primary CPT versus the CPT set with respect to the two filtering conditions on CPT codes. The complementary algorithm improves the results from initial step by 8% on average. Furthermore, the incorporated text features enhanced the quality of the output by 20-35%. Conclusions: We have established a robust framework based on a decision tree predictive model. We predict the surgical codes more accurately and robust compared to the state-of-the-art deep neural structures which can help immensely in both surgery billing and scheduling purposes in such units.

show abstract

Section: Class Weight Recalculation Methodsmentioning

confidence: 99%

Section: Input Datamentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Tree Based Approach for Multi-class Classification of Surgical Procedures Using Structured and Unstructured Data

Khaleghi

Murat

Arslanturk

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Therefore, any prediction attempt directly from unprocessed text results in high prediction error rates. In our prior study, we have identified, reduced, and structured feature sets by utilizing an unsupervised text mining approach from the free-text descriptions [ 5 , 9 ].…”

Section: Introductionmentioning

confidence: 99%

A tree based approach for multi-class classification of surgical procedures using structured and unstructured data

Khaleghi

Murat

Arslanturk

2021

BMC Med Inform Decis Mak

Self Cite

View full text Add to dashboard Cite

Background In surgical department, CPT code assignment has been a complicated manual human effort, that entails significant related knowledge and experience. While there are several studies using CPTs to make predictions in surgical services, literature on predicting CPTs in surgical and other services using text features is very sparse. This study improves the prediction of CPTs by the means of informative features and a novel re-prioritization algorithm. Methods The input data used in this study is composed of both structured and unstructured data. The ground truth labels (CPTs) are obtained from medical coding databases using relative value units which indicates the major operational procedures in each surgery case. In the modeling process, we first utilize Random Forest multi-class classification model to predict the CPT codes. Second, we extract the key information such as label probabilities, feature importance measures, and medical term frequency. Then, the indicated factors are used in a novel algorithm to rearrange the alternative CPT codes in the list of potential candidates based on the calculated weights. Results To evaluate the performance of both phases, prediction and complementary improvement, we report the accuracy scores of multi-class CPT prediction tasks for datasets of 5 key surgery case specialities. The Random Forest model performs the classification task with 74–76% when predicting the primary CPT (accuracy@1) versus the CPT set (accuracy@2) with respect to two filtering conditions on CPT codes. The complementary algorithm improves the results from initial step by 8% on average. Furthermore, the incorporated text features enhanced the quality of the output by 20–35%. The model outperforms the state-of-the-art neural network model with respect to accuracy, precision and recall. Conclusions We have established a robust framework based on a decision tree predictive model. We predict the surgical codes more accurately and robust compared to the state-of-the-art deep neural structures which can help immensely in both surgery billing and scheduling purposes in such units.

show abstract

Intelligent deep learning‐based hierarchical clustering for unstructured text data

Jyothi

Lingamgunta

Eluri

2022

Concurrency and Computation

View full text Add to dashboard Cite

Document clustering is a technique used to split the collection of textual content into clusters or groups. In modern days, generally, the spectral clustering is utilized in machine learning domain. By using a selection of text mining algorithms, the diverse features of unstructured content is captured for ensuing in rich descriptions. The main aim of this article is to enhance a novel unstructured text data clustering by a developed natural language processing technique. The proposed model will undergo three stages, namely, preprocessing, features extraction, and clustering. Initially, the unstructured data is preprocessed by the techniques such as punctuation and stop word removal, stemming, and tokenization. Then, the features are extracted by the word2vector using continuous Bag of Words model and term frequency-inverse document frequency. Then, unstructured features are performed by the hierarchical clustering using the optimizing the cut-off distance by the improved sensing area-based electric fish optimization (FISA-EFO). Tuned deep neural network is used for improving the clustering model, which is proposed by same algorithm. Thus, the results reveal that the model provides better clustering accuracy than other clustering techniques while handling the unstructured text data. K E Y W O R D S fitness improved sensing area-based electric fish optimization, hierarchical clustering, tuned deep neural network, unstructured text data clustering INTRODUCTIONGenerally, speech and text data are read by humans easily, but the machine learning and statistical modeling applications have some unstructured data and so, it is necessary to do some alterations in the coded input feature sets. 1 Data clustering is a technique used for splitting the data elements into many groups so that the elements in the same group have the highest similarity. Though, based on the cluster's attributes, there are diverse elements in other groups. The major aim of clustering techniques is to get centroids or cluster centers for characterizing the entire cluster. Few of the clustering techniques were performed and classified from different scenarios such as "density-based methods, grid-based methods, partitioning methods, and hierarchical methods." 2,3 Moreover, the data set is defined as categorical or numerical. The primary statistical features of numeric data are utilized for describing the distance function between data elements. The categorical data is imitated from the qualitative and quantitative data, and then the descriptions are attained from the counts. 4 By using a "textual virtual schematic model" (TVSM), the textual data are assigned in clusters and it follows three steps. Initially, the extraction of unstructured data is carried out from the data source, and then, it is changed into structured data. 5 After that, clustering is implemented on structured data. Finally, the comparison of documents is done for enhancing the performance of the query based on accuracy.The day today's life generates a huge amount of unstructured text...

show abstract

Automated Surgical Term Clustering: A Text Mining Approach for Unstructured Textual Surgery Descriptions

Cited by 12 publications

References 25 publications

A Tree Based Approach for Multi-class Classification of Surgical Procedures Using Structured and Unstructured Data

A Tree Based Approach for Multi-class Classification of Surgical Procedures Using Structured and Unstructured Data

A tree based approach for multi-class classification of surgical procedures using structured and unstructured data

Intelligent deep learning‐based hierarchical clustering for unstructured text data

Contact Info

Product

Resources

About