The total wait time for patients to receive their medical care has significantly increased in the past few years. The median wait time in 2021 to receive treatment after seeing a family physician and consulting with a specialist is reportedly 175% longer than in 1993.(Moir & Barua, 2021).

Introduction

Having worked as a physical therapist in the USA, I came across many patients having their family members in Canada facing a long wait time to schedule elective surgeries of hip and knee replacements. I always wondered what could be the driving factors contributing to such a challenge within the healthcare system.

Taking a continuing education course "Understanding Canadian Healthcare System" from McMaster University Continuing Education provided me with an opportunity to learn in-depth about the structure and functionality of Canadian healthcare. I researched further on the factors contributing to the wait time and wrote an academic essay on it as a part of the assignment in November 2022.

Here is a link to my assignment essay:

Wait Time For Healthcare

Some of the key findings

Factors contributing to the crisis

• Healthcare Funding Constraints

  Federal healthcare funding constraints cause provincial governments to reduce healthcare resources to operate efficiently and cost-effectively. (Lampkin, 2022).Covid-19 has been one of the factors affecting federal healthcare budgets with provinces having to delay elective surgeries to care for the Covid-19 patients which resulted in further backlogs.

• Healthcare Worker Crisis

  The national healthcare worker crisis has resulted in understaffing situations in healthcare facilities resulting in people having to wait in line at hospitals, emergency departments and long-term care centres to receive the care they need. (Wright, 2022).

• Inappropriate Hospital Bed Assignments

  Chronic national hospital bed shortages and inappropriate bed assignments to the patients due to inadequate funding, lack of system coordination and technical framework in those sectors result in alternative levels of care at the hospital. (Sutherland & Crump, 2013). This further increases wait time as beds are unavailable to care for acute patients.

Potential solutions

• Retain Healthcare Workforce.

  There is an urgent requirement to retain the healthcare workforce in urban and rural communities with incentives of better salaries and working conditions, routine health and mental wellness benefits and increasing child support at work to provide job security for healthcare workers.

• Increase Technological Framework.

  There can be an integrated technological framework with interoperable EHR systems between all healthcare sectors across urban and rural communities and provinces. This integrated system can help in efficient monitoring and coordinating care with a focus on virtual care and e-referrals to timely link patients to the appropriate health providers to reduce wait time.

• Distribute Home-care and Long-term resources adequately.

  There can be increased federal funding to improve home care and long-term care resources with a focus on the proper distribution of those resources between both rural and urban communities. Constructing more publicly funded long-term beds and increasing partnerships with private sectors can help meet the growing elderly population's needs to further reduce wait times.

Conclusion

I enjoyed researching and writing an essay about the wait time and learnt a lot.

I believe that there are many interconnected challenges in addressing this issue despite both the federal and provincial governments taking several funding initiatives and adopting strategies.

There should be major systemic reforms and sustainable long-term change initiatives to be implemented in both the public and private sectors to help Canadians get their timely care.

Thank you for taking the time to read this blog. I would appreciate any feedback or suggestions for improvement.

Resources

Moir, M., & Barua, B. (2021, December 15). Waiting your turn: Wait times for health care in Canada, 2021 report. Retrieved November 2, 2022, from https://www.fraserinstitute.org/studies/waiting-your-turn-wait-times-for-health-care-in-canada-2021

Lampkin, C. (2022, April 08). Federal Budget 2022: Federal Funding for health care fails to meet Canadians' expectations. Retrieved November 2, 2022, from https://www.canadaspremiers.ca/federal-budget-2022-federal-funding-for-health-care-fails-to-meet-canadians-expectations/

Wright, T. (2022, October 31). Canada's ER crisis: Doctors urge governments to stop finger-pointing and find solutions - national. Retrieved November 2, 2022, from https://globalnews.ca/news/9234725/canada-emergency-rooms-solutions/

Sutherland, J., & Crump, R. (2013, August). Alternative level of care: Canada's Hospital Beds, the evidence and options. Retrieved November 2, 2022, from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3999549/

This is an Excel Dashboard project, I completed as a part of practising my Excel skills. I used the stroke dataset from Kaggle to prepare, clean and analyze the data and created a visualization dashboard using different graphs.

You can find the link to my completed Excel project and the stroke dataset here:

Excel Stroke Data Exploration and Visualization

Kaggle Dataset

Data Preparation, Cleaning And Analysis Process

1. I downloaded the CSV file of the stroke dataset from Kaggle and imported it to Excel.

1. I created another sheet duplicating the data to keep the original data intact and use the new sheet for further cleaning and analysis of the data. I further named it the " Working Sheet".

2. I used Sort and Filter features in Excel to get the first look at the data.

3. I started cleaning the data by first checking for duplicate rows but none were found.

4. I increased the visibility of the column labels by using the formatting tools to make the labels bold and change the case of the labels.

5. I found an error in the Work Type column after using a Filter feature which had string value of "Children". I checked with the age column to find they were between 0-16 years of age.

6. I later replaced "Children" in Work Type column to “Unemployed” as a category for clear visualizations.

7. I used the find and replace features in Excel to change the values of 0 and 1 to "Yes" and "No" in the columns of Stroke, Hypertension and Heart Disease for clarity in the visualizations.

8. I created brackets columns for Age and BMI and named them "Age Brackets" and "BMI Brackets" using the following nested "IF" functions. This helped me create clear visualizations of those categories.

=IF(C2>=65,"Senior (65+)",IF(C2>=25,"Adult (25-64)",IF(C2>=24,"Youth ( 15-24)",IF(C2<=14,"Child (0-14)","Invalid"))))

=IF(K2="Unknown","Unknown",IF(K2>=30,"Obese (30+)",IF(K2>=25,"Overweight (25-29.9)",IF(K2>=18.5,"HealthyWeight (18.5-24.9)",IF(K2<18.5,"Underweight (0 -18.4)","Unknown")))))

1. I found and replaced "N/A" values in BMI column with "Unknown" for clarity in the visualizations.

2. I increased the consistency of the strings throughout the columns of Work Type by finding and replacing "Never_worked" and "Self-employed" with "Unemployed" and "Self_employed" respectively.

Key insights from the analysis

1. There are a total of 249 patients who had a stroke with females having higher stroke incidents than males.
   56.7% in females Vs 43.37% in males.

   

   

2. Seniors had the highest percentage of strokes followed by adults and then children.

   

3. People working privately had more stroke incidents than others. Unemployed male individuals had no stroke incidents while unemployed females had 2 stroke incidents.

   

   

4. Individuals who did not smoke had the highest stroke incidents followed by people who formerly smoked and then the ones who smoked.

Limitations in the dataset

1. There could be a sampling bias in the data with a significantly higher number of people who did not have a stroke than the ones who had a stroke.

2. The time frame for which the data was collected is missing which makes it difficult to draw enough patterns in the data.

3. There are a significant number of nulls or missing values which I categorized as unknown.

Conclusion

Skills Used

1. Data cleaning in Excel.

2. Creating categories/ brackets using nested IF functions to use them for visualizations.

3. Creating pivot tables and charts.

4. Creating a visualization dashboard in Excel using customized charts.

5. Inserting sliders to create an interactive dashboard.

My learnings

I enjoyed using Excel for this project as I found it a very useful tool to:

1. Explore the data with some built-in functionalities of sorting and filtering to get the first view of the data, identify the data format and check for errors and duplicates in the data to clean it further.

2. Analyze the data using pivot tables to get useful insights from the data.

3. Create simple and effective visualizations with customized charts.

4. Make the visualizations interactive by using some of its functionalities like inserting multiple slicers.

I appreciate your time to review my project blog and look forward to any feedback or suggestions for improvement.

Resources:

Kaggle Stroke Dataset Prediction

Alex The Analyst YouTube Portfolio Tutorial

Hello all,

I am Namrata Joshi, a Physical Therapist and a NASM Certified Nutrition Coach. I am advancing towards a career change and learning Excel as a part of my journey. I wanted to combine my nutrition knowledge and Excel learnings and hence decided to use this dataset of Nutrition Facts for McDonald's Menu to analyze the data and create a useful dashboard. Any feedback and comments are welcomed and appreciated.

TL;DR

In this analysis, I have found the unique categories of items, menu items with the most proteins and beverages/tea & coffee/ smoothies & shakes low-calorie items with nutrition benefits.

I have used multiple Filters, Sort, Conditional Formatting, and UNIQUE, COUNTIF, SMALL and LARGE functions using various conditions to get to the specific data. I have also inserted different charts to my filtered and sorted data to represent the analysis visually.

The complete analysis can be found in this Google sheet.

Data Analysis

Finding unique menu categories.

I created a copy of the menu in a new sheet and named it menu_dataset. I further applied some formats in the first heading row to get a better read of the data.

Steps

• Creation of distribution of categories sheet

• Use of a UNIQUE function

=UNIQUE(menu_dataset!A:A)

Finding the number of items in each category.

Steps

• I used a COUNTIF function in the second column B.

=COUNTIF(menu_dataset!A:A, ""&A1)

• I used an autofill to copy the formula to the other cells in column B.

• Created a pie chart of unique categories and % of items in them.

Finding the high protein items.

I found the items containing proteins>25 (the approximate amount recommended per meal for muscle protein synthesis). I later calculated % proteins( % of proteins from total calories in an item) and further retrieved a list of the top 5 items with high % protein.

Steps

• Creation of Items with high % protein sheet.

• Finding items with protein>25.

I used the following filter function by selecting a range including categories, items, serving size and calories in menu_dataset sheet with a condition of protein>25.

=FILTER(menu_dataset!A2:D261,menu_dataset!T2:T261>25)

I also used the following filter function in a new column by selecting the range of the protein column in menu_datasetsheet with a condition of protein>25.

=FILTER(menu_dataset!T2:T261,menu_dataset!T2:T261>25)

• Creation of a new column (F)-% protein.

I found out the % of proteins from the total calories of items. It was achieved by getting the total protein calories in an item first (Each gram of protein contains 4 calories) and then dividing it by the total calories of an item.

Hence, I created a new column and used the following formula by multiplying proteins column values by 4 and dividing by calories column values.

=(E2*4)/D2

I used an autofill to copy the formula for the other cells in column F.

• Inserting a note in the heading cell. I inserted the following note in the heading cell- % Protein (F1).

• Highlighting the top 5 items with high % protein. I used conditional formatting and the following LARGE function to achieve that.

=$F2>=LARGE($F$2:$F$38,5)

In the conditional format rule, I selected the range A2:F38 and used format cells if the custom formula is the above function. In this, LARGE function applies conditional formatting to test each row if it is larger than or equal to the 5th largest value. I later custom formatted the style filling the output values with light green 1 colour.

• Sorting the top 5 % Protein items.

I used the following SORTN function where I selected the range($A$2:$F$38)first and selected 5(as the number of results I want), 0 (as only five results to return), 6 (as the nth column to sort), and FALSE (to sort in descending order).

=SORTN($A$2:$F$38,5,0,6,FALSE)

• Creating a chart listing items from highest to lowest % protein. I created a bar graph chart containing the top 5 items having the highest to lowest % protein values

Making healthier choices in menu drinks.

Steps

• Creating a sheet named Beverages, Tea/Coffee, Smoothies/Shakes.

• Finding items meeting their daily recommended values for saturated fats, cholesterol and total sugars.

I tried to find the items in the above categories meeting their daily recommended values for saturated fats<10%, cholesterol< 300 and sugars< 50(Approximate recommended values for total sugars), having some proteins, vitamin A and calcium. For this, I used the following filter function selecting a range of different adjacent and non-adjacent columns meeting the above conditions.

=FILTER({menu_dataset!A112:A261,menu_dataset!B112:B261,menu_dataset!D112:D261,menu_dataset!G112:G261,menu_dataset!P112:P261,menu_dataset!T112:T261,menu_dataset!U112:U261,menu_dataset!W112:W261},menu_dataset!I112:I261<10,menu_dataset!K112:K261<300,menu_dataset!S112:S261<50,menu_dataset!T112:T261<>0,menu_dataset!U112:U261<>0,menu_dataset!W112:W261<>0)

• Finding top 5 less calorie items having nutrition benefits.

I used conditional formatting and the following SMALL function to highlight the top 5 low-calorie items.

=$C2<=SMALL($C$2:$C$16,5)

In the conditional format rule, I selected the range A2:H16, used format cells if the custom formula is the above function. In the formula,SMALL function applies conditional formatting to test each row if it is smaller than or equal to the 5th smallest value. I later custom formatted the style filling the output values with light green 1 colour.

• Sorting 5 low-calorie items in their ascending order using the SORTN function.

a) 5 low-calorie items with % daily value distribution.

This included sorting 5 low-calorie items and their % daily values of Total fat, carbohydrate, Vitamin A and Calcium using the following function.

=SORTN({$A$2:$E$16,$G$2:$H$16},5,0,3,TRUE)

I sorted a few columns with 2 different ranges containing the columns of category, Items, Calories, Total fat(% daily value), Carbohydrate(% daily value), Vitamin A(% daily value) and Calcium( % daily value). I later selected 5(as the number of results I want), 0 (as only 5 results to return), 3 (as the 3rd column of calories to sort), TRUE (to sort in ascending order).

• Bar graph chart

I hid the column of calories and inserted a bar graph chart selecting the columns of Items and (Total fats, carbohydrates, Vitamin A, Calcium) % daily values.

b) 5 low-calorie items with Protein distribution

This includes sorting 5 low-calorie items with their protein values using the following function.

=SORTN({$B$2:$C$16,$F$2:$F$16},5,0,2,TRUE)

I sorted a few columns with 2 different ranges containing the columns of Items, Calories and Protein. I later selected 5(as the number of results I want), 0 (as only 5 results to return), 2 (as the 2nd column of calories to sort), TRUE(sort in ascending order)

• Bar graph chart

I hid the column of Calories and inserted a bar graph chart selecting the columns of Items and Protein.

Lessons Learnt

1. I learnt effective ways to use the FILTER and SORT functions in this large dataset to get to the specific data.

2. I learnt a new way to select a range, especially for the non-adjacent columns which helped me to use the specific columns for the analysis.

3. I found the LARGE and SMALL functions very effective in the conditional formatting to highlight the rows meeting the specific conditions for a better visual representation of the data.

4. I learnt to customize my inserted charts to label the data in the Google sheet.

Resources

Kaggle dataset of Nutrition Facts for McDonald's Menu

Google sheet with detailed analysis

Hello there,

My name is Namrata Joshi, coming from a medical background who practiced physical therapy for almost 12 years. I recently started learning SQL as I am advancing towards a career change and have decided to blog my journey. I found this amazing online source that helped me to not only practice SQL but also learn in an interesting and fun way. I hope this blog comes in handy to anyone who is trying to solve the SQL City Murder Mystery.

Step 1 : Finding crime scene report

I retrieved the crime scene report from the police department’s database with the information which is provided( date= 15th Jan 2018 and city= SQL City, type = Murder). I was able to get the date format by running a query( select * from crime_scene_report ) which helped me to use that in the query below to get a description in the crime_scene_report with the matching criteria.

Query

SELECT * 
from crime_scene_report 
where date="20180115" 
and city = "SQL City";

Output

Step 2 : Finding First Witness

From the above description, there is a clue that there are 2 witnesses. I used the first witness information- “Lives at the last house on Northwestern Dr” to run a query in Person table. This helped me find the name and address_number columns with matching criteria of address_street_name= “Northwestern Dr”. I used ORDER BY to arrange the address_number column data in a descending order( sorting the address_number from last to first) and limited my output result to 3 to get the name of the first result output being the first witness.

Query

Select name, 
address_number
from person
where address_street_name = "Northwestern Dr"
Order by address_number desc
Limit 3;

Output

1st witness- Morty Schapiro

Step 3 : Finding Second Witness

There is another clue in the retrieved crime_scene_report description about the second witness whose name is “Annabel” and lives somewhere on “Franklin Ave”. I used the Person table to find the name and address_number of a person. I used a WHERE Clause inputting the provided information of address_street_name = “Franklin Ave” and used a wildcard %’ later to get a more precise search. For example, using LIKE and % after Annabel to retrieve all the data after the word Annabel. This helped me get the full name of the 2nd witness.

Query

Select name, 
address_number
from person
where address_street_name = "Franklin Ave"
and name like 'Annabel%';

Output

2nd Witness- Annabel Miller

Step 4: Finding the witnesses interview transcripts

After getting the full names of the first and 2nd witnesses, it was important to find their interview transcripts individually from the interview table. Hence, I queried both the Interview and Person tables as the transcript was needed from the interview table and the name had to be used in the person table. I used a WHERE clause to join two tables using the primary keyid from the person table and foreign key person_id.) and used an AND clause to input the name of each witness in individual queries to get their interview transcripts.

1) Morty Schapiro

Query

  select person.name, 
  interview.transcript
  from person, interview
  where person.id = interview.person_id
  and person.name = 'Morty Schapiro';

Output

2) Annabel Miller

Query

select person.name, 
interview.transcript
from person, interview
where person.id = interview.person_id
and person.name = 'Annabel Miller';

Output

Step 5: Finding Gym Members

I combined the interview transcript information of both the witnesses as they were related. Both said that the murderer was recognised as the "Get Fit Now Gym member". Hence I had to join two tables and use a get_fit_now_member table to get the name, id and membership_status information. I used the provided information of check_in_date from get_fit_now_check_in table and membership_id starting with “48Z” from get_fit_now_member table. I added the WHERE clause to join the two tables using the primary key id from get_fit_now_member table and the foreign key of membership_id from get_fit_now_check_in table. I further applied the AND’clause and a wildcard % to input the exact information provided by the witnesses. This helped me get a precise output data on names of gold membership status members with their membership ids starting with “48Z”.

Query

select get_fit_now_member.name, 
get_fit_now_member.id,
get_fit_now_member.membership_status
from get_fit_now_member, get_fit_now_check_in
where get_fit_now_member.id  = get_fit_now_check_in.membership_id
and (get_fit_now_check_in.check_in_date = '20180109' 
and get_fit_now_member.id like '48Z%');

Output

Step 6: Querying the drivers_license table

Using the license plate information provided by "Morty Schapiro", I wanted to find the name of the person having a car license plate number containing “H42W”. For this, I used the two tables of person and drivers_license. I joined two tables using a WHERE clause matching the primary key of id from drivers_license table and foreign key of license_id from person table to retrieve the common data present in both tables. I used an AND clause and a wildcard % after “H42W” to get the list of names and their respective drivers license plate numbers containing the input data.

Query

Select person.name,
drivers_license.plate_number
from person, drivers_license
Where person.license_id= drivers_license.id
and drivers_license.plate_number like '%H42W%';

Output

Step 7: We found the murderer

The murderer is “Jeremy Bowers” after merging the outputs of step 5 and step 6.

Step 8: Verification

a) Checking "Jeremy Bowers" interview transcript.

I was able to find the interview transcript of Jeremy Bowers using both person and interview tables. I used those two tables asid and name columns were present in person tables and transcript column was present in the interview table. I joined those tables using the primary key of id in the person table and foreign key of person_id in the interview table with a WHERE clause. I used an AND clause to specify name specifically being ''Jeremy Bowers''.

Query

Select person.id, 
person.name,
interview.transcript
from person,interview
where person.id = interview.person_id
and person.name= 'Jeremy Bowers';

Output

b) Finding the brain behind the murder

Option 1

I had to find the name of the person and the person id matching the transcript description provided by Jeremy Bowers. I used these 4 tables of person, drivers_license, income and facebook_event_checkin in the query below. I joined the tables using their keys with a WHERE and multiple AND clauses. I input the information of hair color being red, car make being Tesla and gender being female to retrieve specific data matching those information.

Query

select person.id,
person.name
from person,drivers_license, income, facebook_event_checkin
where person.license_id = drivers_license.id
and 
(person.id  = facebook_event_checkin.person_id 
and person.ssn = income.ssn and drivers_license.hair_color= 'red'
and drivers_license.car_make= 'Tesla' 
and drivers_license.gender = 'female');

Output

Option 2

I used a subquery first to find the person's id, person’s name, height, car model, income from 3 tables of person, drivers_license and income. I also included a WHERE and multiple AND clauses to join the three tables matching their keys and input the hair color, car make and gender information as derived from Jeremy’s interview transcript. I later used an ORDER BY to sort the annual_income column data in the descending order. I later ran a query to find nameand id of the person from the above subquery as well as from the facebook_event_checkin table. It was important to join the personand facebook_event_checkin tables using keys of id and person_id respectively to retrieve the matching data.

Query

select id, name 
from 
       (select person.id,
        person.name,
        drivers_license.height,
        drivers_license.car_model,
        income.annual_income
        from person,drivers_license, income
        where person.license_id = drivers_license.id
        and (person.ssn = income.ssn 
              and drivers_license.hair_color= 'red'
              and drivers_license.car_make= 'Tesla' 
              and drivers_license.gender = 'female')
        Order by income.annual_income desc)
,facebook_event_checkin 
where id = facebook_event_checkin.person_id;

Output

Mystery Solved

Brain Behind the murder

Miranda Priestly

Lessons Learnt

1. It was helpful to first query all the different tables to get a better read of the inside data.
2. I learnt to think creatively to join different tables depending upon the information I was getting from each query.
3. It was fun to play around with a few wildcards and filters to retrieve the exact data although it took some trials and errors to get rid of the few syntax errors.
4. I could also have used a combination of different JOINS in my queries although it was easier to filter the data using WHERE and AND clauses.
5. Readability was an issue in some of my subqueries until I started to add some indentations.
6. It is always helpful to think about multiple ways to retrieve the required data.

Resources

SQL City Murder Mystery