Data in motion: IBM's Big data analytics case study
- ali@fuzzywireless.com
- Mar 4, 2022
- 2 min read
The case study for data in-motion is picked from IBM’s big data and analytics hub and located at below URL titled as “Real-time analysis for intensive care”:
After registering at IBM’s big data and analytics hub, the article can be downloaded from below hyperlink:
Data in-motion referred to the analysis of big data in real-time or near real-time without storing first (Ebbers, 2013). Case study of data in-motion here is the real time analysis of vital signs from neonatal intensive care unit specifically for premature newborn babies (Blount et al., 2010). Premature babies can acquire infections from intensive care unit which are known as Nosocomial infections. Nosocomial infections are difficult to diagnose because of vague, nonspecific and subtle signs using traditional clinical practices. Other dominant infections in premature babies are Pneumothorax, Intraventricular hemorrhage and Periventricular Leukomalacia which are analyzed in real-time along with Nosocomial infection to detect and intervene earlier and reduce morbidity and infant mortality.
Framework by Blount et al. (2010) collects and analyze the physiological streaming data of ECG, heart rate, respiratory rate and blood SpO2 in real-time. The collected streaming data is than compared against the established patterns of infections and trigger diagnosis for early intervention. Security and privacy measures were taken to safeguard the personal health data under the privacy laws of Canada, United States and province of Ontario. Model was deployed successfully in neonatal intensive care (NICU) of the SickKids in Toronto and accurately identified signatures of infections in early stages.
Core of the framework is the stream computing system which is capable of processing high velocity and high volume streaming data (Blount et al., 2010). Stream processing application declarative engine (SPADE) is used to develop the framework. SPADE can be used for existing as well as new pattern correlation data of an infection on the patient’s streaming physiological data for diagnosis. Different clinical rules can be applied for instance if mean arterial blood pressure is less than 24 mmHg for 20s and if SpO2 is less than 85% for the same 20s period then trigger an alert.
Data in-motion analysis works on the logic of analyze first and store later and can be applied in cases where time sensitive decision making is critical, for instance:
1. Flight traffic control system
2. Stock market
3. Interpretation of physiological data for medical intervention
4. Network fault detection and remediation
Reference:
Ebbers, M. (2013) 5 Things to know about big data in motion. Retrieved from https://www.ibm.com/developerworks/community/blogs/5things/entry/5_things_to_know_about_big_data_in_motion?lang=en
Blount, M., Ebling, M., Eklund, J., James, A., McGregor, C., Percival, N., Smith, K & Sow, D (2010) Real-time analysis for intensive care. Retrieved from http://public.dhe.ibm.com/software/data/sw-library/infosphere/whitepapers/Real-Time-Analysis-for-Intensive-Care.pdf
Comments