Presentation on theme: " E-Health Patient Records Analysis By Gian Frez (el13gcf) and Matthew Hughes (ed10m2jh)"— Presentation transcript:
E-Health Patient Records Analysis By Gian Frez (el13gcf) and Matthew Hughes (ed10m2jh)
E-Health Patient Records Overview Electronic health records (EHRs) are electronic health data recorded by health care professionals, as well as patients, including: Medical history and medication records Daily charting, diagnosis and test results Nursing notes and care plan Purpose: To increase the availability of information across healthcare systems/practices To improve the efficiency and effectiveness of the healthcare system To support research into new diseases and intervention trials NHS Connecting for Health (UK), the Personally Controlled Electronic Health Record (Australia), e-Health Exchange (USA)
Case Study Data and Text Mining the Electronic Medical Record to Improve Care and to Lower Costs 1 1 P. Cerrito and J.C. Cerrito, “Data and Text Mining the Electronic Medical Record to Improve Care and to Lower Costs,” in Proc. SUGI 31, 2006, paper 077-31.
Data Description Examine EHRs to determine the treatment of patients in the Emergency Department (ED) EHRs over a 6-month period were analysed using SAS software Data contains triage information, final outcome (patient disposition), medication and treatment time
Medication Clusters All medications for each patient were transposed and concatenated into one text string SAS Text Miner defined clusters of medication treatments, resulting in 13 different clusters
Final Disposition and Medications Relationship between final disposition and medications Cluster 7 has the highest proportion of home discharges Clusters 3 and 13 have a majority of patients admitted to hospital
Transactional Time Series Analysis SAS PROC HPF used to examine patient time series data 3,300 patient visits to ED were examined over 3 months Data accumulated by the hour and then averaged to determine changes over a 24-hour period
Average Treatment Time Average treatment time (LOS) over 24 hours:
Explanation of Results The peak in treatment time was due to an increase in the number of patient visits Statistically significant, r 2 = 74%
High Performance Forecasting (HPF) HPF to determine whether there is a regular and predictable pattern in patient treatment time (LOS) Optimal prediction was seasonal/periodic
Conclusion EHRs can be mined and analysed to improve the healthcare system e.g. the ED SAS Text Miner compared final disposition and medications SAS PROC HPF used to examine patient time series data Analyses can be extended e.g. tracking demand for other hospital services to optimise facility use, develop protocols, improve scheduling and reduce waiting time
References Denny JC (2012) Chapter 13: Mining Electronic Health Records in the Genomics Era. PLoS Comput Biol 8(12): e1002823. doi:10.1371/journal.pcbi.1002823 Blog.withings.com. How e-health data mining can help scientific research. | Withings blog. [Online] Available from: http://blog.withings.com/en/2012/04/05/how-e-health-data-mining- can-help-scientific-research/ [Accessed 30 Nov 2013]. Research.microsoft.com. eHealth - Microsoft Research. [Online] Available from: http://research.microsoft.com/en-us/collaboration/global/asia- pacific/programs/ehealth.aspx [Accessed 30 Nov 2013]. P. Cerrito and J.C. Cerrito, “Data and Text Mining the Electronic Medical Record to Improve Care and to Lower Costs,” in Proc. SUGI 31, 2006, paper 077-31.