HADM 250 Data Anytc in Health SVR Ad MN Notes (April 6-17 2020) PDF

Preview

Summary

This document contains all of my lecture notes, screenshots, and discussion post writings for the first two lectures given in April of 2020....

Description

Lecture Notes 4/6/2020 Objectives ● ●

Navigation and Visualization Capabilities ○ Demonstration of Application Presentation Introduction to Structured Process ○ PEER Planner

-The whole idea of data analytics is a team effort, its not just an individual

Dr Akram Boutros (COO of South Nassau Communities Hospital) "We have to bring the science of management back into healthcare" Berwick MD Institute of Healthcare Improvement He could only deal with the politics in DC for 18 months before he bailed out Agenda ●

●

● ● ● ●

Competitive Consideration: ○ Market Trends ○ Geographic Mapping & Assumotions PEER PLANNNG Forecasts: ○ Dashboard Update ○ Long term predictions ○ Competitive Module Physician Activity: ○ Responsible for admission and care of the patient so its ultimately up to them Operational Impact: Continue to see more closures and consolidations, more focus on physician activity, how many do we have? How many do we need? Can also measure operational activity

There's a lot of analytics that go into building the dashboard After the dashboards, we can get into competitive simulation modeling

Can do model fitting Can see some high peak value that represents occupied beds Capacity of the hospital was 340 beds SNCH Forecast We see the best case scenario is 22k patients a year, 3 years later that’s what the hospital did generate

● ● ● ●

Looking at cost through the window of cost charges Want to identify nurses we need to have to support these patients (simulation of what we may be trying to look for) We can break this up in a lot of ways, we can begin to see the trends this represents Button allows us to convert all of this to graphs to better understand what this actually looks like

● ● ●

What goes into putting all of this together? The numbers are never absolute, the statistical noise is +/- something The difference in the number of days in a month contributes to the statistical noise ○ Ex. January has 31 days that contributes to 10% of the difference

● ●

Intermittency means there's a 0 and we don't want that We want the data to be big enough to show homogeneity consistently

Independent Discussion Post The article Understanding the COVID-19 I thought was interesting but I could not help thinking a couple of things. The first thing, it would have been more beneficial for these organizations to have these tools to better understand disease from different angles and make predictions based on the data they have. The problem is, with all the tools they want to implement it will take time for these organizations to gain access to it and then acquire the data they actually need to analyze to see if there are any patterns they can unlock. The second thing, because this is rapidly growing pandemic, it is hard to get reliable up to date data (all of the data assembled together) without some kind of margin of error to predict the direction COVID-19 is going especially because all of the tools haven't been implemented yet. I do find it interesting however that artificial intelligence can streamline targets to get closer to a vaccine. Implementing this successfully I believe is all dependent on what data is shared and accessible to get the most out of the data relating to COVID-19. Response to Classmates Post When a patient is engaged in their healthcare, he can partake in making healthcare decisions, but most importantly understands his health regime and can abide by it. The patient’s health experience is based on the quality, efficiency, and effort placed forth by the provider at every episode of care. Patients should be regarded as persons and empowered consumers who have ample choices of places to receive their health services, so health organizations should take heed of their given feedback. The article discusses the use and importance of available data sources, the quality of the data collected and how that data could be used to predict the patients’ behavior. It mentions various surveys from which allows providers assess which patient/customer interactions need more emphasis, namely: ● patient/provider communication, ease access to services (website views and/or wait for an appointment w/provider), address issues, percentage of call resolution that didn’t require f/up ● patient’s loyalty- family referrals, feedback, overall satisfaction, rate of appointment bookings versus demand, patient satisfaction Although those surveys provide somewhat an in-depth analysis of patients’ satisfaction and touchpoints in the form of data points; I believe that special emphasis should also be placed on the engagement of diverse populations. Too often, those patients become a financial loss due to providers’ lack of cultural competency. Providers should always be prepared to serve their growing community of patients (including foreigners trapped here because of COVID19) by

integrating those performance measures to improve its service delivery. Healthcare providers need to adhere to the healthcare standards by providing culturally and linguistically appropriate services (CLAS) to those patients requiring language access services. Only then would the data collected be a true reflection of its impact on its patients’ outcomes and their contribution to effectively improve patients’ experiences, health literacy, and health equity. I agree with your point that providers should be seen in that light and more engagement with diverse populations. More cities, states and neighborhoods are becoming integrated so that changes the needs of the population the more diversified it comes so its worth putting some more focus on. However in this situation, since availability for care is scarce, patients are going any and everywhere to receive care if they are affected by COVID-19 so certain areas will be integrated by default. There are still barriers that need to be broken such as the language barrier as you mentioned. It is already a scary thing to know that our healthcare delivery is already underprepared as far as supplies and tools, but what they can do is prepare in the best way possible to serve the people they are not use to serving and putting people in place to ensure the best communication and care delivery possible.

Discussion Board Individual Post

IV Overcoming Interoperability Challenges I agreed with this portion of the articles point that the healthcare industry has overcome many obstacles relating to interoperability, sharing meaningful data between organizations and providers but it does currently have its limitations which can cause major problems. I thought the percentage of successful sharing data between health systems was quite shocking (only 38% success rate) and it just shows how far the healthcare industry has come from not sharing any data at all but also how far we still need to go to share vital data to serve patients better. For example, the limitations we are still figuring out how to overcome is affecting the reliability of the data relating to COVID-19. Not only does the public have limited if not any access, but the organizations are still not getting up to date information that they can analyze and make valuable so that we are well informed of what is going on around us and how healthcare facilities can respond. I personally believe once nationwide exchange of EHI is implemented successfully, sharing data will be a lot smoother. Classmate Post Looking back on the past 10 years since HITECT Act was passed we’ve come to see there are better outcomes we can strive for. Physician burnout, interoperability hassles and

implementation delays are just some of the missed opportunities HITECT now has to focus on. Improving the system would focus on enhancing electronic health data, driving physician communication which can be the biggest struggle for the healthcare system as we are all competing against one another. Incorporating more HIT professionals, realizing that tech is here to stay, and healthcare has become a complex environment. Lastly working together to over interoperable challenges in the workplace. HITECH offers a vary of solutions for clinical workflow, it improves patient care and team collaboration. This is important given what are county is facing now, big data is always relevant and increasing the use of HITECH can help elevate most of our challenges.

Response to Classmate Danielle, I wholeheartedly agree with the points you made. I also believe it is imperative to add tech staff to help implement these systems to analyze data such as electronic health records for patients and not the healthcare providers themselves. Having physicians and other providers put a good portion of their focus into data related matters like implementing systems where they practice acts as a very big deterrent for them. It keeps them from doing their jobs, and their priority is caring for the patient. It is not their fault because they are required to do these things as of right now, but this is one of the many changes that needs to be made as soon as possible. I believe by taking these steps, it will be one factor that will help reduce the level of physician burnout, care delivery will be a lot smoother, forecasting with data will run a lot better, the data will be the most up to date for analysts to work with so they can go through proper procedures to make sense of the data even with the statistical noise.

4.17.20 Lecture Notes Announcements: On the syllabus where it says there was originally going to be a presentation along with the analysis paper due before May 4th, there will be No presentation due to the environment

Lecture 8 ●

Its important to recognize the quality and timeliness of the data is very important

●

PEER Planner Forecasting-

Objectives Slide ● Prepare data, execute results, evaluate what we are looking for, and reconcile what we believe are the differences.

● ●

●

There is a repetitive cycle of what we go through week after week in order to update our forecast. Have to evaluate the quality of the results and make adjustments. If new PC's are added to the hospital for example we can reconcile the numbers to make them more realistic We can be looking at quarterly, monthly or weekly data and we have to look for seasonality

Summary Level Considerations ●

Every single product both, cancer, pulmonary they all have a degree of seasonality

●

Generating the Baseline Forecast

●

If you were to click on Forecast model automatic, it would Calculate all9 methodologies

State Space Forecasting Models ●

For this model, you would have to have at least three years of history in order to effectively work with these

State-Space Forecasting Models (Univariate Family of Exponential Smoothing & ARIMA) ● ● ● ●

SPF is a univariable forecasting approach for statistically extrapolating historical patterns (trend and seasonality) over a forecast horizon SPF consists of a powerful family of exponential smoothing techniques, each appropriate for a particular pattern Can automatically and optimally forecast a wide variety of data having trend and multiplicative/additive seasonal patterns Has multiplicative error distribution leading to asymmetrical (NEW) and usual symmetrical prediction limits

Seasonal Exponential Smoothing Models ● ● ● ●

Referred to as Holt-Winters model Allows for additive or multiplicative seasonality Allows for additive or multiplicative error distribution - nontraditional asymmetrical prediction limits Methods works by ○ Starting at the current level ○ Adding the product of the current trend by the number of periods ahead we are projecting

○

Adjusting the resulting sum of level and trend for seasonality with an additive or multiplicative seasonal index

Workshop Demonstrations of PEER Planner *Not in slides*

●

You can refine your search by different categories, time frames (in years), type of

● ●

condition etc. Depending on what type of records and how many of them, calculations can be done in a manner of minutes *During live demonstration* The professor's search and calculations took a little over sixty seconds to complete...