Skip to course
HARVEY CASTRO, MDDR GPT™ LABS

PHYSICIAN LEADERSHIP IN THE AI ERA

Ward to Board

Bring a real workflow.
Leave with a better way to evaluate it.

A twelve week course with Harvey Castro, MD. Measure the work, question AI claims, plan a safe test and teach what you learn.

Course preview · Enrollment is closed

Educational use only · No CME credit · Course disclosures

12 weeksThree phases with a working document each week

Written lessonsReview sheets, exercises and linked evidence

Professional developmentNo accredited CME or clinical competency claim

IS THIS USEFUL FOR YOUR WORK?

Start with a decision you face.

Designed for physicians and clinical leaders who need to evaluate AI in a real workflow. Bring a question about documentation, verification, team impact or patient communication.

A place to begin

You can work from a fictional example or a workflow you are authorized to examine. Coding and buying a tool are not required for the written activities.

A usable record

Build a baseline, a verification checklist and a small project plan. A documented decision to pause can be the right result.

Created by Harvey Castro, MD. Read his background and work.

THE CURRICULUM

From observations to a decision.

Open a week to see what you will work on. The course also includes a welcome module and a source guide.

01Map the real work

Record where time goes. Separate active work, waiting and interruptions.

Your working document: A time and energy map.

02Examine your assumptions

Name a belief about AI and test it against evidence and your actual experience.

Your working document: A belief audit.

03Protect meaningful work

Use your own stories to describe what matters without treating reflection as a cure for burnout.

Your working document: A personal decision statement.

04Measure one workflow

Record a baseline, identify safeguards and choose a primary and a balancing measure.

Your working document: A workflow map and baseline.

05Test a documentation task

Use an approved task and check the complete workload, including review and corrections.

Your working document: A bounded trial record.

06Verify before relying

Check sources, assumptions and outputs. A fluent explanation does not establish accuracy.

Your working document: A verification checklist.

07Plan with your team

Ask who gains time, who takes on work and whose role changes. Record concerns in their own words.

Your working document: A transition and training plan.

08Build patient trust

Prepare an explanation consistent with the actual use, jurisdiction and institutional policy.

Your working document: A disclosure and consent preparation sheet.

09Scope an achievable test

Plan a small approved test for the four remaining weeks. Schedule longer work separately.

Your working document: A project scope and approval map.

10Document the safety case

Trace the data, name the accountable owner and define monitoring and stop conditions.

Your working document: A safety case and monitoring sheet.

11Plan for the real quarter

Choose one featured action, a recognizable cue and one fallback. Review and adjust without a streak penalty.

Your working document: A practical plan for the next quarter.

12Teach and check application

Explain what you observed and ask a colleague to apply a verification rule to a new case.

Your working document: A team teaching session and follow up plan.

READ A COMPLETE SAMPLE · WEEK 11

A plan that fits your actual week.

Choose a commitment, name likely obstacles and decide how you will review it. Keep other obligations visible and adjust as your week changes.

Open the sample lesson

The next quarter will include clinical work, interruptions and days when your capacity is lower. Build a plan that can change with those conditions.

Choose a direction, then one action

Choose one meaningful outcome for the next ninety days. Write how you will observe it, what is within your control, and who else must agree. You may keep other outcomes on your worksheet, but choose one action to focus on this week.

This is a course design choice to keep the review manageable. Research does not establish that every physician can hold exactly three commitments, or that one plan is universally optimal.

An outcome might be a shorter referral turnaround time. An action might be reviewing two approved, deidentified examples with the workflow owner. Keep the primary measure and a balancing measure, such as errors or work transferred to someone else, together.

Put the action at a recognizable moment

Use this sentence:

When [specific situation], I will [small action].

For example: When my Thursday administrative block starts, I will review two referral examples against the verification checklist.

Then add one fallback:

If [likely obstacle], I will [smaller action or a new time].

For example: If the block is interrupted, I will review one example in Friday's protected time. If that time is no longer available, I will pause and choose again at the weekly review.

Use a situation you can recognize. A plan that says "work harder when things settle down" gives you little help deciding what to do next. The cue is a reminder, not permission to interrupt patient care.1

Make the first step easy to find

James Clear recommends reducing the setup steps between a person and a useful action. This is a practical design suggestion; it does not guarantee that the action will happen.2

Original course application. Put the blank verification checklist and the approved fictional example in the same folder before your chosen work block. At the cue, open that folder. If access or approval is missing, resolve it before proceeding. Do not remove a safeguard to make starting easier.

A short weekly review

Choose one recurring window. Start with an eight minute budget and adjust it to your needs. Eight minutes is a practical starting point, not a proven dose for physicians.3

  1. Did the cue occur? If you did not record it, mark it unknown.
  2. What happened: completed, interrupted, canceled, rescheduled or not recorded?
  3. What does the measure show, including errors, added work and missing observations?
  4. Keep the action, shift its time, reduce its scope or pause it.

A missed week is information. There is no automatic second week failure rule and no broken streak penalty. Write the reason so that the next plan reflects your actual conditions.

Make room safely

Look for work you can stop, delegate or reduce with the agreement of its owner. Do not drop a clinical safeguard, contractual duty or another person's work without checking the consequences.

If nothing can safely move, shrink or pause the project. Protect sleep and recovery. A course should not require you to create time that you do not have.

When estimating a recurring task, use recorded durations from comparable completed instances if available. Separate hands on time from elapsed time. Mark interruptions and missing records rather than filling gaps from memory. Historical timings can inform a forecast; they cannot remove future uncertainty.

Ask for help if it helps

You can invite a colleague to review the plan with you. Agree what you want to share, when you will talk and what kind of support is useful. An accountability partner is optional. This course does not promise that naming one guarantees follow through.

Your reflections remain your original words. Share a summary only if you choose to. No message should be sent on your behalf without your review.

Review the quarter

Ninety days is a planning window, not a claim about how long habits take to form. At the end, compare observations using the same definitions. Record what you kept, changed or stopped and why. A decision not to continue can be the appropriate result.

For the live session

Bring your calendar, one proposed action, its cue and its fallback. Ask a partner to identify a likely obstacle and check that your response preserves clinical responsibilities and recovery. Leave with a plan you have chosen, not a quota of commitments.

Sources used in this material

  1. Sheeran P, Listrom O, Gollwitzer PM. The when and how of planning: Meta-analysis of the scope and components of implementation intentions in 642 tests. European Review of Social Psychology; online 2024, print 2025. Publication bias adjustment and limits of moderator comparisons. Research synthesis. Effects vary and do not establish a universal optimal number of plans or a physician specific benefit. ↩

  2. James Clear. How to Improve Your Health and Productivity Without Thinking. Author article, undated; accessed November 20, 2026. How to Design Your Environment for Success. Author advice. Only the practical suggestion to reduce setup steps is used. No health effect or automatic adherence is claimed. ↩

  3. Uhlig L and colleagues. A field experiment on the effects of weekly planning behaviour on work engagement, unfinished tasks, rumination, and cognitive flexibility. Journal of Occupational and Organizational Psychology, 2023. Methods and Table 2. Field experiment. Working adults, not a physician cohort. Median task time is not a proven prescribed dose. ↩

Read the evidence and limits

Implementation intentions

Sheeran P, Listrom O, Gollwitzer PM. The when and how of planning. European Review of Social Psychology. Online 2024, print 2025. DOI and primary article.

This synthesis covers 642 tests. Estimated effects vary with the analysis and adjustment for publication bias. The robust Bayesian estimate was d = 0.15, with a 95 percent interval of 0.08 to 0.22. This is not a physician specific effect or a forecast for an individual learner. Cross study comparisons do not establish an optimal number of plans. In particular, the paper does not justify a universal limit of three, or prove that one plan is always better.

Weekly planning

Uhlig L and colleagues. A field experiment on the effects of weekly planning behaviour on work engagement, unfinished tasks, rumination, and cognitive flexibility. Journal of Occupational and Organizational Psychology, 2023. Primary article.

The study followed 208 working adults with 947 weekly observations. Its planning prompt had several components. The median completion time was 7.9 minutes. Table 2 reports lower unfinished tasks and rumination, higher cognitive flexibility, and no clear effect on engagement. It does not isolate an eight minute prescription or establish results in physicians. Associations in the mediation analysis should not be presented as the only causal mechanism.

What the course chooses

One featured action, a fallback, a short review and neutral rescheduling are design decisions to test for usefulness and burden. Ninety days is an administrative review interval. A named partner is optional; no comparative superiority or guaranteed adherence is claimed. Check whether the plan helps you, including time spent reviewing it.

Evidence reviewed November 20, 2026. The original supplied research reports are retained unchanged in the audit archive.

Source register

The numbered credits in the lecture take you to its references and back. These records identify the sources used this week. The evidence notes above explain findings and limitations. No book purchase is required.

Course application

The worksheet, fictional cases and suggested timing are course design choices. They have not been validated as a clinical assessment or as a guarantee of outcomes. No named author has endorsed this course.

Optional further reading

The sources above are enough for this activity. Additional reading is optional.

Open the printable planning worksheet

This blank worksheet is for your own notes. The preview does not collect or save entries.

Plan for the next quarter

Harvey Castro, MD · Ward to Board

This is a planning aid, not a validated assessment or a habit formation timetable.

Start date: Review date:

Direction

One outcome that matters to me:

Primary measure and baseline:

Balancing measure (errors, burden or work transferred):

What needs another person's agreement:

This week's commitment

When [specific situation], I will [small action]:

If [likely obstacle], I will [fallback]:

What can safely make room for this:

If nothing can move, how I will reduce or pause the project:

Prepare the first step

Setup idea from James Clear: reduce unnecessary steps before a useful action. This is an original course application of author advice, not an adherence guarantee.1

The approved example and checklist I will place together:

What must be authorized before I begin:

Weekly review

My chosen day and time:

Start with eight minutes; adjust if needed. Do not count missing records as failures.

Week Did the cue occur? Outcome What I observed Keep / shift / reduce / pause
1 yes / no / unknown
2 yes / no / unknown
3 yes / no / unknown
4 yes / no / unknown
5 yes / no / unknown
6 yes / no / unknown
7 yes / no / unknown
8 yes / no / unknown
9 yes / no / unknown
10 yes / no / unknown
11 yes / no / unknown
12 yes / no / unknown
13 yes / no / unknown

Outcome options: completed, interrupted, canceled, rescheduled, not recorded.

Optional colleague check in and what I choose to share:

Quarter review

What changed, using the same measurement definitions:

Missing observations or other explanations:

What I will keep, change or stop:

Sources used in this material

  1. James Clear. How to Improve Your Health and Productivity Without Thinking. Author article, undated; accessed November 20, 2026. How to Design Your Environment for Success. Author advice. Only the practical suggestion to reduce setup steps is used. No health effect or automatic adherence is claimed. ↩

TRY THE METHOD

Would you repeat this claim?

A fictional example. Choose a response, then compare your reasoning with the explanation.

An impressive number without a method

A vendor says its documentation tool saves 40 percent of the work. Its slide does not identify the sample, comparison or time spent correcting drafts. What is the best next step?

What would you do next?

This selection is not saved. It is a practice exercise, not a validated clinical assessment.

EVIDENCE WITH ITS LIMITS

Know what the course can claim.

A source behind the statement

Lessons link findings to their population and setting. A management metaphor, small pilot or vendor report is not treated as a universal rule.

Practice before confidence

Short cases ask you to choose before reading feedback. Completion, self reported confidence, case performance and later recall remain separate.

Time choices to evaluate

The twelve week sequence, biweekly practice and brief weekly review are course design choices. They are not established optimal doses for physicians.

Your institution's requirements

Clinical changes need the appropriate approvals. The course does not authorize patient data use, replace governance or promise a clinical outcome.

Content review: November 20, 2026. Written materials are available. Recorded lectures, live dates and paid enrollment are not available in this review edition.

BEFORE YOU START

The format and the limits.

What is available to read now?

This public page includes the curriculum, a full sample lecture, its evidence notes, a printable worksheet and one practice case. The protected owner review contains twelve written lectures, welcome and resources, thirteen worksheets and six practice cases.

How much time should I set aside?

Each lecture gives an estimated reading time and a suggested exercise duration. Review the exercise before choosing a time. The twelve week sequence is a course design choice; it is not a requirement to work through illness, inadequate sleep or clinical responsibilities.

Do I need to use real patient information?

No. Use fictional, nonidentifying examples in the course. Do not upload patient or confidential institutional information. Any use of real information or a change in clinical care requires the appropriate local approvals.

Are there videos, live sessions or CME credit?

This review edition contains written material. Recorded lectures and confirmed live session dates are not included, and no accredited CME credit is offered. The fictional practice cases are ungraded and do not certify clinical competence.

Can I enroll, and what does it cost?

Enrollment is closed, and a paid offer is not available on this page. Format, dates, support, terms and price need to be confirmed before enrollment opens. You can ask about the course without making a purchase.

Where are my answers and notes saved?

Practice choices are temporary and are not saved to a learner account. The review page does not collect personal notes. Download a worksheet to keep your own copy in an appropriate private location. Read the course disclosures for the review page's privacy behavior.

START WITH ONE REAL QUESTION

What would you need to know
before changing this workflow?

That question carries through the course, from your first baseline to the conversation with your team.

Return to the course mapAsk about the course
Course scope and disclosures

Review edition · November 20, 2026

Purpose and professional judgment

Ward to Board teaches methods for evaluating healthcare workflows and AI claims. It does not provide individual medical care, diagnosis, treatment or legal advice. It does not replace professional judgment, current clinical guidance or your institution's requirements.

Use of AI and patient information

Use fictional material for course exercises. Do not enter patient information, confidential institutional data or another person's private information into this review website or its practice activities. Downloading a worksheet does not make it an approved patient data system.

A lesson, checklist or correct practice answer does not authorize a tool or a clinical change. Verify the actual output and obtain the approvals required for the intended use, data and setting. Responsibility and legal obligations depend on the facts and applicable rules.

Evidence, accuracy and AI assistance

AI assistance was used in drafting and editing this review edition. Author and independent clinical review remain in progress. The evidence notes distinguish research findings, limitations and course design choices; they do not establish that every statement is error free.

Check the original source, its date and its relevance before relying on a finding. Laws, institutional policies and product capabilities can change. Statements about jurisdiction specific requirements require qualified review for the actual setting. Mentioning a product or study does not establish its suitability for a particular patient or institution.

Practice, outcomes and CME

The six practice cases are fictional and ungraded. Their feedback is educational, not a validated assessment of clinical competence. Completion, confidence, application and delayed recall are different observations.

This review edition offers no accredited CME credit or clinical credential. It does not guarantee time savings, reduced burnout, career advancement, improved patient outcomes or any other result.

What the review page saves

The course review page does not save your practice answers or personal notes to a learner account. Practice choices are temporary within the open page and reset when you change material or leave it. Downloads remain on your device. The owner review code protects access to the full course; it does not make this a patient information workspace.

The website's privacy policy applies to its general operation. Opening the full owner review stores a short lived session in the current browser tab. This review mode is separate from any future enrolled learner classroom, which will need its own verified privacy and data handling information before launch.

What is available now

This edition contains written lectures, evidence notes, review sheets, worksheets and practice cases. Enrollment is closed. Recorded lectures, confirmed live session dates and paid learner services are not included. These scope disclosures are not a reviewed enrollment contract or a blanket liability waiver.

Website privacy policy

Return to the course ↑