![User Avatar](https://mu-informatics.org/wp-content/uploads/learn-press-profile/380/1e64e18b0049bcae65ef50652a10badc.jpeg)
After learning about the research, we have seen several opportunities in big health data. On the other hand, it has brought quite challenging issues that still need to be tackled to improve patient outcomes. Some top challenging issues describing the paper are missing data, selection bias, data analysis and training, and privacy issues.
Missing data: Those seriously impact analysis and give invalid results depending on the number of missing data. Several factors have contributed to generating missingness in the databases. Imputation techniques are one of the models that can correct missing data recommended in the paper. Another suggestion is to provide regular refresher training to staff for data validation so that a certain number of missing data can be reduced at each level.
Selection Bias: The risk of selection bias due to the inclusion of subjects from different geographic, insurance and medical history profiles were compared in this large-scale EHR analysis. Recording sampling methods and being transparent in reporting for selection bias in the analysis is recommended.
Data analysis and Training: EHR usually entails a large dataset to analyse multiple times to hypothesise an event’s significance eventually. It applies algorithms several times to handle the complexity. The government has to invest more in the sector to develop skilled staff in order to support the industry and provide regular training and workshops to widen the knowledge and skills of the staff in the statistical area.
Privacy and ethical issue: People who share their health data have the right to protect their privacy. Cybercrimes target private information to hack for several reasons and advantages. Setting up a robust security system is one of the solutions to defend hackers from unauthorised access to big health data. Sensitive information should be encrypted before it is shared with designated research.