Page 250 - Big Data Analytics for Intelligent Healthcare Management

9.5 ADVANTAGES AND DISADVANTAGES 243

Health record providers

Hospitals
Medical researchers

Community health staff
Medical clinics Centralized
server

Doctors

Patients
Patients’ records
HRS
(Health recommendation system)

FIG. 9.6
Health recommendation system (HRS) architecture.

Table 9.3 Comparison Among Existing Approaches and Proposed HRS
Contribution Average MAE Values (Patients) Average MAE Values (Doctors)
Yakut and Polat 0.724 0.795
Kaur et al. 0.739 0.807
Proposed HRS 0.649 0.717

In Table 9.3, MAE values are shown for the healthcare dataset where 10,000 patient ratings for 500
doctors are divided among 5 parties. Here p represents a number of parties collaborating, which varies
from 1 to 5 where p¼1 meaning that there is no collaboration and all parties are generating predictions
individually. The p¼2 signifies that two parties are collaborating and likewise for p¼3, 4, or 5.
Fig. 9.7 depicts that MAE value is lower when all parties are collaborating, that is, p¼5. The lower
the MAE value, the higher the accuracy. By using a collaborative-based filtering technique on the pro-
posed HRS, we achieve lower MAE values and high accuracy when compared to existing approaches.

9.5 ADVANTAGES AND DISADVANTAGES OF THE PROPOSED HEALTH
RECOMMENDATION SYSTEM USING BIG DATA ANALYTICS
The proposed HRS has some advantages. Real-time remote tracking of vital signs through embedded
sensors (attached to patients) allows healthcare providers to be alerted in case of anomalies in pulse
rate, heartbeat, disorientation, etc. Big data analytics help doctors to get quick access to health