Puneet Vohra

So I am taking the example of Health care domain:
Background:
In Health insurance companies the claim processing systems often denies certain Medical procedures due to missing or unclear prior authorization details. These denials triggers the Manual Appeals process, where the Appeal resolving analyst handles  review and either uphold or overturn the denials based upon the additional medical documents submitted by providers or members.
 
Repeated Human exceptions:
Analysts notice that a particular type of CPT code for spinal injections (ex -99213) was being frequently denied due to missing prior authorization, , but during the Appeals : The provider is categorically indicating that is a urgent Need and submit in the clinical notes as well Authorization had been requested but delayed The procedure was performed in an emergency setting, exempting it from pre-authorization.  
Impact - More than 80 % of the cases were reversed the denial on Appeal after reviewing the additional notes/data.
 
Human in the loop learning loop:
Initial AI performance  The ML/AI based machine will flag the appeals with missing Prior authorizations and valid denials, offering no recommendation for reversal Human interventions to modify the AI generated outcomes: Analysts consistently Overturned(Approved) those cases which contains  structured Comments such as mentioned below: " Emergency setting, Prior Auth not required policy X"
"Provide er submitted auth request but payer systems delayed it"
Pattern Identification: The AI system uses NLP(natural language processing) to read clinical notes and Appeal attachments/documents. Its resembles the keywords such as Clinical urgency phrases, emergency settings , or Auth request timestamps with Overturned decisions Model Re-training:  AI model is required to retrained periodically with these type of examples  which eventually results in:
Achieve the  probabilities of that a denial will be overturned Suggest to Auto escalate for Appeal or even can Pre-approve under certain conditions Impact:
 Reduction in unnecessary manual Appeals by 30 % Min.  Faster resolution times for Providers and patients Improved trust from Providers due to fewer incorrect denials

Yes! While writing the problem definition in Define phase of DMAIC. There can be different opinions of different stakeholders. Here are some  practical steps which can be utilized to avoid initiating a project with misunderstood problem:
 
Root Cause Analysis:
Use Techniques such as 5 Whys or Fishbone Diagram to identify  the root causes of the issues by ensuring that each stakeholders actively participate in this analysis to provide their input Data Collection:
Gather the volume number and defects data as well to understand the problem with the lens of data Facilitate workshops:
Encourage the people to speak their opinions on the problem and group their thoughts based upon similarity Problem statement Drafting:
Draft a problem statement based on the insights gathered Ensure that problem statement is Specific , measurable achievable relevant and timebound Validation and Consent  buying from stakeholders:
Present the drafted problem statement to all stakeholders for validation Encourage feedback and make necessary adjustments to ensure everyone agrees on the problem definition Documentation:
Document the agreed-upon problem statement and ensure it is accessible to all stakeholders Include the logic and supporting documents of problem definition To avoid misunderstood Problem:
Do a transparent and open communication throughout the Define Phase Regular Check ins Be prepared to refine the problem statement  iteratively based upon the new insights or feedback

Gantt Charts provides a visual timeline, it makes easy to see the sequence of activities, the durations and dependencies. Its helps in breaking down the project in to manageable tasks, assigning responsibilities and setting deadlines.   When we update the chart project managers can monitor progress and recognize delays and make necessary adjustments It serves as communication tool helping stakeholders  to understand the project status. How can we overcome the limitation of Gantt chart, while we are working on complex dynamic projects :
We have to ensure the Gantt chart is continuously updated to reflect the current status of the project Use Gantt charts in conjunction with other project management tools like Kanban boards, task lists and Agile methodologies to provide a more comprehensive view  Be ready to change the Gantt charts as the projects evolves. This reevaluating task durations, dependencies and resource allocation.  Incorporate risk management practices to predict and resolve challenges that are impacting the project timeline  Regularly communicating with stakeholders to ensure they are aware of changes and can provide support as required  focusing on key milestones rather than indulging in the details of every task. This helps is maintaining the high- level view of the project's progress.

The below answer is portrayed considering the 'Health insurance and medical billing' process to draw a comparison between standard work and work instructions:
 
 
Aspect
Standard Work
Work Instructions
Definition
The group of activities illustrating the best way to handle claim submission, medical billing and denial management
It comprises the detailed explanation of how to perform each and every process/step within the medical billing and insurance process
Scope
Broadly it covers steps such as claim submission, coding, insurance verification and payment posting
It covers detailed steps such as claims form filling, claim uploading in portal, claim receiving in clearing house, claim processing and adjudication, payment posting , letter sending
Purpose
The purpose of standard work is to standardize the process for all teams ensuring consistent handling of claims, appeals and other workflow activities
It provides the guidance in terms of specific process steps and compliance guidelines too with payer rules
Level of detail
L1 level steps emphasizes on activities such as verifying insurance, coding, submitting and reviewing denials and submitting them to payers etc.
L4 and L5 level which provides click level information of each process and also includes the screenshots and images to facilitate the reader.
Example
Workflow for claim submission including verifying insurance, claim adjudication, Claim error verification, coding submission by medical biller of doctor and reviewing denials
Instructions of coding claims such as ICD - 10 and CPT codes, it includes specific payer rules and software usage guidance also
Purpose in process
Standard work is a practise to ensure the process which is free of error and accelerating the claims processing
It ensures the accurate execution  of specific tasks such claim form filling , claim verification, policy coverage verification, claim validation and payment processing and letter generation
Audience
To get the high-level view of the process for same set of practices to be followed by all team members
The team which follows the work instruction need to know the payer rules, application and software use for case handling
 
 
 

'Workflow Analysis plays a crucial role in shaping the DMADV framework:
 
Define phase: Work flow analysis helps to understand closely the As is flow of the business process.
Example: If a health insurance company is going to develop a super-fast claim processing and resolution system that has been analyzed by seeing the As Is flow that claims resolutions are often delayed Then the problem definition in business case would be reducing processing time by 20%.
 
Measure phase: If the workflow analysis is done during this phase that helps in collecting data of actual business metrics, also this data is utilized as benchmark for the new process
Example: If the claim resolution of treatment and hospitalization expenses is taking 30 days. This is our Measurement baselining. The Core team working on projects can aim to reduce the claim resolution time by 50 %.
 
Analyze Phase: In Analyze phase, workflow analysis is done to recognize the waste of entire process from the lens of 'People, Process, Technology and Control'
Example: An analysis might show that manual claim form adjudication or processing is a bottle neck process. This insight can lead team to design of an automated claim form processing by leveraging the OCR and AI to create the automated data entry which will impact the efficiency of the process.
 
Design phase: In design phase the outcome of workflow analysis can help us to re-write the 'To be flow' and develop the solution considering the recognized gaps and needs.
Example:   If the hospital or patient want that they should get the real time update of claim processing as well then the solution which is designed must have the automated notification sending feature incorporated.
Verify Phase:  The workflow analysis helps to test the designed solution as per the Business requirement , its makes sure the 'To be process' meets customer expectations.
Example: Once the new claim processing automated solution is 'Go live'. The core members can utilize the new process reduces the resolution time and improves the customer satisfaction of the claimant.
 
Conclusion: By integrating the workflow analysis into the DMADV framework, the risks are minimized and increases the probability successful deployment of solution in operations to resolve and process the claims of expenses done by the hospital and patients which ultimately reduces turn around time.

Tap root Analysis - Tap root analysis and Root cause analysis is the same thing used when we work on any issue/problem. The ultimate moto is to find the root cause of the problem. It is used very often by quality and process excellence professionals to improve the business. In this analysis the problem is seen from each lens i.e. People, Process, Control and Technology and then under each lens the points are listed. Once this is done then the classification of all causes is done in to the 'Data door' and 'Process Door'
 
Steps:
Recognize, understand and define the problem. Collect the relevant data and details pertaining to that recognized issue. Make plan which comprises of sequence of activities leading up to the problem this helps in understanding where things went wrong. Develop a chart that visually maps the various factors that contributed to the incident. Make use of Fishbone diagram, Why-Why Analysis, Fault Tree Analysis to dig deeper in to the reasons behind the failure To offer and develop the solutions or actions which eliminates the root cause ensuring these actions are feasible and sustainable.  To sustain the implemented solution by continuously monitoring initially for 1- 2 months ensure that problem does not recur.  
 
Benefits and limitations of Tap root analysis, Root cause and Why-Why Analysis:
 
 
Benefits
Limitations
Root Cause Analysis
Ideal for team brainstorming sessions to dig the problem
For Manufacturing Industry - Causes are classified in Man, Machine , Material , Method
For Service Industry -  Causes are classified in to the People, Process, Control, Technology
Encourages the collaboration with teams of upstream and downstream process
Identifies probable causes however does not dig in to the each one.
Does not provide the sequence of activities leading to the identified problem
Tap Root Analysis
Systematic approach for comprehensive analysis
 
 
 
Provides multiple causes
Easy representation of causes through charts of complex problems
Prevention oriented
Complex issues and highly technical problems
Due to its depth takes longer time
Complex required specific training to execute the analysis
Why-Why Analysis
No specialized training, very simple approach
Provides only one root cause.
 
Can be  done very fast for smaller problems
Very much subjective
 
Focuses on cause and effect relationship by deliberately asking 'why'
Not favourable for highly complex issues
 
Tap Root analysis are better suited for below mentioned cases:
 Cyber security breach in Finance domain  Safety Bypass in Manufacturing industry  Surgical errors in Healthcare domain  Hose failure in High speed racing cars  Aviation Safety issues in aircrafts  Space crafts Helium gas leakage  Railway accidents  Gas leakage from the chemical processing plant

Skewness is a way to explain how the data is spread out around the mean. It tells us whether the data is more falling on one side or the other rather than being distributed equally
 
What does it indicate?
 
Direction of Tilt- Skewness reflects if the data is more towards the right , If yes then its is (Positive skewness) or it is towards left then its is (Negative Skewness)
 
Where is our mostly data - It shows where the most data points are positioned. If its is s case of Positive Skewness then data points are on the lower side. In negative Skewness- Data points are on the higher side
 
Outliers- Skewness can also suggest if there are unusual values or outliers that are far from the rest of the data
 
Types of Skewness
 
Positive Skewness (Right- Skewed)- Here the Mean is moreover larger than the median because the higher value pulls it up
Example= A batch of 20 candidates participated in Master black belt program . Most candidates scored  between 70% -80% out of 100 %, only some of them obtained  95% or more. The point here is that Mean score will be coming 85% due to the high scores but most students actually scored around 70 - 80 %, which makes the Mean higher than the typical for the batch . As a result of which we might think the class did good overall than they actually did
 
Negative Skewness (Left- Skewed)- Here mostly data points are on the higher side, however some low values pulls the average down.
Example- Think of a small company where mostly people are getting 20000 - 40000 INR salary, whoever is working from long back older is earning 10000 INR. The point here is that Mean salary will be coming around 25000 INR because of some old people getting 10000 INR salary. Even though the mostly people are getting 30000 INR salary. As a result of which mostly people will look getting less salary than the actually get.
 
Zero Skewness(- The distribution of datapoints is equally distributed)
Example when we play Ludo and imagine that the dice is giving 3,4,5,6 repetitively for number so times we played. In this case Mean will be 4.5 and mode will also be 4 or 5. The point here is that dataset will be evenly distributed actually each number has come about the same number of times
 
How it affects our Interpretation?
Skewness is crucial to understand for correct data interpretation ensuring that statistical analyses give valid and reliable result. It also represents that datasets are evenly spread or not that actually impacts the mean, mode and median and other measures