Page 63 - Big Data Analytics for Intelligent Healthcare Management

REFERENCES 55

(e) For analytical purposes, the dynamic availability of the different machine learning algorithms or
the extended ones of the earlier algorithms to work on such large scale of data should be easily
accessible such as a pull-down type menu.
(f) Concurrency in big data analytics should be maintained in an efficient way so that data
inconsistency should not occur at any instant or at any cost as otherwise this will lead to a
serious problem of affecting the whole healthcare industry concerned.

The above are some inevitable challenges that should be addressed appropriately for efficient and
optimal big data analytics in healthcare.

3.8 CONCLUSION AND FUTURE RESEARCH
Adoption of big data technology is rapidly increasing in the healthcare industry. The medical im-
aging field, which may be considered as an automation process of the manual diagnosis process,
is also dependent directly or partially on medical big data since an accurate diagnosis of a serious
disease at the right stage needs continuous study and research on a huge volume of diagnosis data
collected from patients with similar symptoms from different clinics from different geolocations.
For efficient big data analytics in healthcare data, there should be a standard framework or model
through which an optimal result might be expected. Also, for implementation, we need to select the
right platform and tools. Besides this, there are several other challenges that need to be addressed
throughout the analysis phase. All these issues are discussed in this paper. Although big data ana-
lytics in healthcare has great potential, the discussed challenges need to be addressed and solved to
make it successful.
For future research, these challenges will be focused on and a novel framework will be built to in-
clude all the necessary steps for accurate medical big data analysis. At first, an empirical study will be
conducted to investigate the techniques that can be adopted for the raw medical big data cleansing and
normalization process. As the medical datasets are a mixture of structured, semistructured, and unstruc-
tured data, the data transformation step will be given more importance. And as a big portion of medical
big datasets is contributed to by medical images, which are generally of a fuzzy nature, to develop an
advanced fuzzy set-based technique for such medical image data enhancement is an unavoidable aspect
of the future research. After the preprocessing step, the remaining portion of the big data analytics will
be more similar to traditional big data analytics, so, future research will mainly target the mandatory
preprocessing steps.

REFERENCES
[1] R. Zhang, H. Wang, R. Tewari, G. Schmidt, D. Kakrania, Big data for medical image analysis: A performance
study, in: Parallel and Distributed Processing Symposium Workshops, 2016 IEEE International, IEEE, 2016,
pp. 1660–1664.
[2] D. Charles, M. Gabriel, M.F. Furukawa, Adoption of electronic health record systems among US non-federal
acute care hospitals: 2008-2012, ONC Data Brief 9 (2013) 1–9.
[3] L. Wang, C.A. Alexander, Big data in medical applications and health care, AME Med. J. 6 (1) (2015) 1.
[4] E. Dumbill, Making Sense of Big Data, 2013.