SELECT gender, org_name
FROM PATIENTS
INNER JOIN microbiologyevents
ON patients.subject_id = microbiologyevents.subject_id
WHERE org_name = "ESCHERICHIA COLI" AND gender IN("F", "M")Analysis Report Two - Data, Data Everywhere
Executive Summary
The healthcare landscape has evolved substantially from paper charts to data being collected from individual glucose monitors and sent back to the physician. Overall, the sheer amount of data that is found in a healthcare system database is constantly increasing. Being able to decipher and mine this data is vital to the improvement of healthcare. (Sutton et al. (2020)) Utilizing collected data, artificial intelligence and other computer models can help detect issues on mammograms that a radiologist might have missed and currently in the dental world can better detect which teeth have a cavity. Overall, the future equity of this data is to help patients and practitioners make quicker, faster, cost-friendly decisions about how to treat the various ailments patients have. (Arowoogun et al. (2024))
Introduction
Data isn’t much accept a bunch of random codes to the most healthcare workers. But, to data analyst it is a mine where they create queries to extract data to possibly change the face of healthcare. (Karatas et al. (2022)) Utilizing this data programs can be built that help guide practitioners through steps of a diagnosis pathway that coincides with current guideline-approved therapy. (Karatas et al. (2022)) This is a prime example of knowledge-based clinical decision support systems (CDSS). (Sutton et al. (2020)) On the other side of the spectrum there are non-knowledge-based CDSS who relies more of artificial intelligence, computer algorithms, associations, recognition of patterns in treatment, etc. to create recommendations to help providers make a decision about a patient’s care. (Sutton et al. (2020)) The ability to combine the massive power of artificial intelligence with that of a physician or other provider is the future of healthcare (Krishnamoorthy, Dua, and Gupta (2023)). In order for physicians to better diagnose, treat, and prevent various diseases these tools are going to be necessary to not only achieve healthier outcomes but also decrease cost to the patient and healthcare system.
The Healthcare Context
As with any new, innovative device technology, specifically data driven CDSS, will have to prove its value in the healthcare setting. (Mamlin and Tierney (2016)) Drugs for one have to go through extensive research and development pathways before they are able to come to market. (Tamimi and Ellis (2009)) Not to mention, there is a lag in practitioner use of those new therapies. Utilization of clinical decision systems that are built from the years of healthcare data will face a similar lag but, the utilization will be much quicker when compared to new drugs. (Mamlin and Tierney (2016)) When executed correctly these systems will allow practitioners to better diagnose, treat, and prevent many diseases that constantly increase healthcare spending like heart failure. These systems will also be a great resource for new practitioners as they develop their own clinical judgement when they start treating patients.
However, as with any technological system it is human nature to come to rely more on the technology than their own brains. (Patil, Myers, and Lu-Myers (2025)) These CDSS are meant to be a tool, not replace practitioners, or replace the need for practitioners to be educated thoroughly. As with some healthcare programs now there is integration with guidelines that groups patients by what specific criteria they meet (i.e. all diabetics are grouped together with the clinical message of needing a cholesterol medication). Many patients with chronic disease often complain about the amount of medications they have to take and when the practitioner is adding on more due to CDSS recommending it this could lead to a problem. The problem of non-adherence by the patient. (Bosworth et al. (2011)) Practitioners can prescribe medications to patients but they may never take 1 dose of that medication. This is one of the largest challenges facing healthcare today.
With the invention of “MyChart”, data integration among healthcare systems has become much easier. With this ease of data sharing comes the problem of formating. Each healthcare practitioner has different formats for progress and treatment notes and each healthcare system has different levels of access that they are willing to share with other organizations. So, data integration is a great thing if you are able to actually access and read data input by another system. This leads to decreased quality of care as practitioners often have to order extra labwork, scans, etc. because they cannot interpret data from another visit at a different facility. This is especially a problem with outpatient data translating to inpatient data. (Redelmeier and Kraus (2018))
Visualization One - Two Table Join
ggplot(data = myquery1,
mapping = aes(x = org_name, fill = gender)) +
geom_bar() +
theme_minimal() +
labs(
title = "Visualization showing number of Male & Female ICU patients w/ ESCHERICHIA COLI",
subtitle = "Data taken from PATIENTS and MICROBIOLOGYEVENTS table within MIMIC-III",
x = "Organism Name",
y = "Number of Men & Women w/ ESCHERICHIA COLI",
caption = "Source: MIMIC-III Clinical Database v1.4"
)
This visualization expresses the amount of patients admitted to the ICU who had a blood test result (i.e. blood culture) come back positive for the bacteria, E. Coli. This visualization also shows the amount of men versus women who had this bacteria during their ICU stay. This information is relevant in healthcare because it helps healthcare systems and practitioners better understand the likely microorganisms they may see causing infections in future patients, as well as, whether males or females are more acceptable to the pathogen. This data helps lead to the critical decisions making of selecting the correct antibiotics in a timely manner in order to reduce patient stay length and increase patient outcomes.
Visualization Two - Three Table Join
SELECT CAST(dose_val_rx AS INT) dose, marital_status, diagnosis, drug, drug_type
FROM PRESCRIPTIONS
INNER JOIN PATIENTS
ON PRESCRIPTIONS.subject_id = PATIENTS.subject_id
INNER JOIN ADMISSIONS
ON ADMISSIONS.subject_id = patients.subject_ID
WHERE marital_status Like "married" AND diagnosis LIKE "pneumonia%"
AND drug LIKE "vanco%" AND dose > 250ggplot(data = myquery2,
aes(x = marital_status, y = dose)) +
geom_violin(fill = "green") +
theme_minimal() +
labs(
title = "Married ICU patients with Pneumonia and Given Vancomycin > 250 mg",
subtitle = "Data taken from PATIENTS, PRESCRIPTIONS, and ADMISSIONS table within MIMIC-III",
x = "Marital Status",
y = "Vancomycin Dose",
caption = "Source: MIMIC-III Clinical Database v1.4"
)
This visualization provides data focusing on the antibiotic, Vancomyin. This is a popular drug of choice in many ICU settings and this visualization shows the amount of married patients diagnosed with pneumonia who received doses greater than 250 mg. This is important in the healthcare setting for one on the inventory side where you know how much of the drug has been used so you know how much you need to order. Secondly, this visualization helps healthcare systems better understand the prescribing habits of the practitioners. Allowing the system to perform evaluations of drug appropriateness and dose appropriateness
Recommendations for Industry
Utilizing clinical decision support systems (CDSS) is the future of healthcare. Whether this is a knowledge-based or non-knowledge based sytem, these systems will continue to be integrated into the various healthcare delivery softwares used by healthcare systems, practitioners, and patients. These systems will help to increase the quality of care individuals receive and help reduce other healthcare problems down the road.
In terms of defensive data management, the integration of CDSS will increase the demand for protective mechanisms for the data. Especially for the data that will be shared to other healthcare databases. This forms the need for the appropriate encryption and maximum security to ensure laws, like HIPAA, are followed. Since healthcare organizations already utilize a lot of data protection mechanisms this enhanced protection will likely evolve as the CDSS evolves.
In terms of offensive data management, the use of CDSS will help to provide healthcare systems to help their practitioners provide better care for their patients. Artificial intelligence can already better detect various things on an x-ray compared to a radiologist. That point along is a major win for CDSS use in healthcare. Additionally, when combined with practitioner knowledge these systems grow even stronger and provide healthcare systems with data that can be analyzed to help reduce hospital re-admissions or hospital-acquired infections. The possibilities are endless of CDSS if implemented appropriately and respected as a tool and not a replacement for a human’s clinical judgement.
References
Arowoogun, Jeremiah Olawumi, Oloruntoba Babawarun, Rawlings Chidi, Adekunle Oyeyemi Adeniyi, and Chioma Anthonia Okolo. 2024. “A Comprehensive Review of Data Analytics in Healthcare Management: Leveraging Big Data for Decision-Making.” World Journal of Advanced Research and Reviews 21 (2): 1810–21.
Bosworth, Hayden B, Bradi B Granger, Phil Mendys, Ralph Brindis, Rebecca Burkholder, Susan M Czajkowski, Jodi G Daniel, et al. 2011. “Medication Adherence: A Call for Action.” American Heart Journal 162 (3): 412–24.
Karatas, Mumtaz, Levent Eriskin, Muhammet Deveci, Dragan Pamucar, and Harish Garg. 2022. “Big Data for Healthcare Industry 4.0: Applications, Challenges and Future Perspectives.” Expert Systems with Applications 200: 116912.
Krishnamoorthy, Sreelakshmi, Amit Dua, and Shashank Gupta. 2023. “Role of Emerging Technologies in Future IoT-Driven Healthcare 4.0 Technologies: A Survey, Current Challenges and Future Directions.” Journal of Ambient Intelligence and Humanized Computing 14 (1): 361–407.
Mamlin, Burke W, and William M Tierney. 2016. “The Promise of Information and Communication Technology in Healthcare: Extracting Value from the Chaos.” The American Journal of the Medical Sciences 351 (1): 59–68.
Patil, Shefali V, Christopher G Myers, and Yemeng Lu-Myers. 2025. “Calibrating AI Reliance—a Physician’s Superhuman Dilemma.” In JAMA Health Forum, 6:e250106–6. 3. American Medical Association.
Redelmeier, Donald A, and Nicole C Kraus. 2018. “Patterns in Patient Access and Utilization of Online Medical Records: Analysis of MyChart.” Journal of Medical Internet Research 20 (2): e43.
Sutton, Reed T, David Pincock, Daniel C Baumgart, Daniel C Sadowski, Richard N Fedorak, and Karen I Kroeker. 2020. “An Overview of Clinical Decision Support Systems: Benefits, Risks, and Strategies for Success.” NPJ Digital Medicine 3 (1): 17.
Tamimi, Nihad AM, and Peter Ellis. 2009. “Drug Development: From Concept to Marketing!” Nephron Clinical Practice 113 (3): c125–31.