Use of AI in Evidence-Based Plan Management: Case Study 2 Fatigue and Sleep Disturbance (Major Depressive Disorder)

Week 9 Assignment: Use of AI in Evidence-Based Plan Management (Case Study 2: Fatigue and Sleep Disturbance)

Use an AI tool (ChatGPT, Grok or another platform) as a learning partner to develop an evidence-based management plan. Include 3–5 differential diagnoses with rationale; a final diagnosis supported by subjective, objective and pertinent history findings; a problem statement; an evidence-based management plan (diagnostics, medications with complete prescriptions, referrals/consultations, patient education and follow-up); a reflection on the accuracy, usefulness and limitations of AI; and an appendix with your AI prompts and outputs. Support your work with 3–5 peer-reviewed references from the last five years.

Case: “I can’t sleep and feel exhausted all the time.” A 45-year-old male reports 6 months of fatigue, poor concentration and difficulty sleeping. He feels “tired but wired,” often lying awake for hours. Appetite decreased; 8-lb unintentional weight loss. Feels “hopeless”; work stress overwhelming. Denies suicidal ideation, hallucinations or substance use. PMH: hypertension, well controlled on lisinopril. FH: father had depression in his 50s. SH: divorced, works in IT, lives alone, 1–2 beers on weekends, no tobacco or drugs, limited exercise. ROS positive: low mood, insomnia, poor appetite, fatigue, decreased motivation; negative: chest pain, SOB, suicidal thoughts, hallucinations. Objective: BP 122/80, HR 84, afebrile; mildly disheveled, flat affect, minimal eye contact; soft, slowed speech; depressed, congruent affect; A&O ×3, memory intact. PHQ-9: 19. GAD-7: 6. Meets five or more DSM-5 criteria for major depressive disorder for longer than two weeks.

Original example for study purposes. Doses follow standard references; always confirm with current prescribing information.

Differential Diagnoses

1. Persistent depressive disorder (dysthymia). Shares chronic low mood, low energy, poor concentration and hopelessness. Ruled out: requires depressed mood for at least two years; symptoms have lasted six months and are severe enough to meet full major depressive episode criteria.
2. Adjustment disorder with depressed mood. Work stress is an identifiable stressor. Ruled out: adjustment disorder is diagnosed only when criteria for another mental disorder are not met; he meets full MDD criteria, and symptoms have persisted for six months.
3. Hypothyroidism (depressive disorder due to another medical condition). Fatigue, low mood and poor concentration overlap. Less likely: weight loss rather than gain, no cold intolerance, constipation or bradycardia (HR 84). Must be excluded with TSH before final confirmation.
4. Bipolar disorder, current episode depressed. Important to exclude before starting an antidepressant. Less likely: no reported history of elevated mood, decreased need for sleep with high energy or impulsivity; “tired but wired” reflects insomnia with fatigue, not mania. Screen with the Mood Disorder Questionnaire and ask about prior episodes.
5. Generalized anxiety disorder / insomnia disorder. “Tired but wired” and sleep-onset difficulty suggest hyperarousal. Ruled out as primary: GAD-7 of 6 indicates mild anxiety only, and insomnia occurs as part of the depressive episode. Insomnia will still be treated as a target symptom. Unintentional weight loss also warrants screening for medical causes (see diagnostics).

Final Diagnosis

Major depressive disorder, single episode, moderate-to-severe, without psychotic features (F32.1 if moderate; F32.2 if severe on full assessment). He reports at least five DSM-5-TR symptoms for more than two weeks, including depressed mood and loss of interest (decreased motivation), plus insomnia, fatigue, decreased appetite with 8-lb weight loss, poor concentration and hopelessness, with functional impact at work. Objective findings support the diagnosis: flat, congruent depressed affect, psychomotor slowing (soft, slowed speech), poor self-care (mildly disheveled) and minimal eye contact. A PHQ-9 of 19 falls in the moderately severe range (15–19). The family history of depression in his father increases risk. Specifiers are pending exclusion of medical causes, substance-related causes and any history of mania or hypomania. Suicide risk is low at present (denies ideation), but risk factors include male sex, living alone, divorce, hopelessness and alcohol use.

Problem Statement

A 45-year-old divorced male who works in IT and lives alone presents with a 6-month history of depressed mood, decreased motivation, sleep-onset insomnia with feeling “tired but wired,” fatigue, poor concentration, decreased appetite, 8-lb unintentional weight loss and hopelessness in the setting of work stress. He denies suicidal ideation, hallucinations, tobacco and drug use and reports 1–2 beers on weekends. History is significant for hypertension controlled on lisinopril and paternal depression. Examination shows normal vital signs (BP 122/80, HR 84, afebrile), mildly disheveled appearance, flat and congruent depressed affect, minimal eye contact, soft and slowed speech and intact orientation and memory. PHQ-9 is 19 (moderately severe) and GAD-7 is 6 (mild). Negative for chest pain, dyspnea and psychotic symptoms.

Evidence-Based Management Plan

Diagnostics

• TSH to exclude thyroid disease; CBC for anemia; CMP for electrolytes, glucose, renal and hepatic function before prescribing and to evaluate weight loss; vitamin B12 and folate; HbA1c.
• Alcohol screen (AUDIT-C) and urine drug screen if history is uncertain.
• Columbia Suicide Severity Rating Scale (C-SSRS) screener at baseline and each visit; Mood Disorder Questionnaire to screen for bipolar disorder.
• Age-appropriate cancer screening: unintentional weight loss at 45 should prompt confirmation that colorectal cancer screening is up to date; further workup if weight loss continues despite improved mood.
• STOP-BANG questionnaire if snoring or witnessed apneas are reported.

Pharmacologic treatment

For adults with moderate to severe MDD, the American College of Physicians recommends either cognitive behavioral therapy or a second-generation antidepressant as initial treatment, and suggests the combination may be considered (Qaseem et al., 2023). Antidepressant choice should be guided by side effects, comorbidities and patient preference (Simon et al., 2024). Mirtazapine is selected because its sedating and appetite-stimulating effects target his prominent insomnia and weight loss, it has few sexual side effects and it has no clinically significant interaction with lisinopril.

Hypnotics such as zolpidem and benzodiazepines are avoided as first-line options because of dependence, falls and next-day impairment, and because he lives alone and drinks alcohol.

Referrals and consultations

• Psychotherapy: cognitive behavioral therapy (CBT) for depression, with CBT for insomnia (CBT-I) components, because combined treatment improves outcomes.
• Psychiatry if there is no adequate response after two adequate antidepressant trials, if bipolar features emerge or if suicide risk increases.
• Employee assistance program for work stress and short-term counseling.
• Primary care coordination for weight loss workup and hypertension follow-up.

Patient education

• Diagnosis: depression is a common, treatable medical condition, not a personal weakness; it affects sleep, appetite, energy and concentration.
• Medication: how and when to take it, expected timeline, side effects and the importance of not stopping suddenly.
• Lifestyle: regular sleep and wake times, limiting screens and caffeine late in the day, daily physical activity such as 20–30 minutes of walking, regular meals, avoiding alcohol and reconnecting with supportive people.
• Red flags: thoughts of self-harm or suicide, feeling unable to stay safe, new agitation or unusually high energy, fever or sore throat. Call or text 988 (Suicide and Crisis Lifeline), call 911 or go to the nearest emergency department.
• Safety planning: identify warning signs, coping strategies and contacts; discuss limiting access to lethal means.

Follow-up

• 1 week: phone or telehealth check for side effects, sleep and safety.
• 2–4 weeks: in-person visit; repeat PHQ-9 and C-SSRS; review labs; consider increasing to 30 mg if response is limited.
• 6–8 weeks: assess response (at least 50% PHQ-9 reduction) and remission (PHQ-9 below 5); if inadequate, optimize the dose, switch or augment, and strengthen psychotherapy.
• Every 4–12 weeks after response; continue treatment 6–12 months after remission.
• Emergency precautions: go to the emergency department or call 988 or 911 for suicidal thoughts or inability to care for himself.

Reflection on the Use of AI

The AI tool was useful for quickly generating a broad differential, organizing the plan into the required sections and suggesting patient education language. It was mostly accurate on DSM-5-TR criteria and first-line treatment classes. However, it made several errors that required correction. It labeled a PHQ-9 of 19 as “severe,” although the validated cutoffs place 15–19 in the moderately severe range. It initially recommended zolpidem for insomnia without discussing dependence, alcohol interaction or fall risk. It omitted the TSH and suicide risk tools until prompted, did not address the unintentional weight loss as a separate problem and gave a prescription without quantity or refills. Its references included one article that I could not locate, which shows that AI can fabricate citations. These limitations match concerns in the literature about the accuracy, bias and accountability of large language models in clinical care (Haltaufderheide & Ranisch, 2024). I verified every recommendation against current guidelines and drug references, and I did not enter any identifying information. AI was a helpful learning partner, but the final plan reflects my clinical judgment.

References

Haltaufderheide, J., & Ranisch, R. (2024). The ethics of ChatGPT in medicine and healthcare: A systematic review on large language models (LLMs). npj Digital Medicine, 7, Article 183. https://doi.org/10.1038/s41746-024-01157-x

Qaseem, A., Owens, D. K., Etxeandia-Ikobaltzeta, I., Tufte, J., Cross, J. T., Jr., Wilt, T. J., & Clinical Guidelines Committee of the American College of Physicians. (2023). Nonpharmacologic and pharmacologic treatments of adults in the acute phase of major depressive disorder: A living clinical guideline from the American College of Physicians. Annals of Internal Medicine, 176(2), 239–252. https://doi.org/10.7326/M22-2056

Simon, G. E., Moise, N., & Mohr, D. C. (2024). Management of depression in adults: A review. JAMA, 332(2), 141–152. https://doi.org/10.1001/jama.2024.5756

US Preventive Services Task Force, Barry, M. J., Nicholson, W. K., Silverstein, M., Chelmow, D., Coker, T. R., Davidson, K. W., Davis, E. M., Donahue, K. E., Jaén, C. R., Li, L., Ogedegbe, G., Pbert, L., Rao, G., Ruiz, J. M., Stevermer, J., Tsevat, J., Underwood, S. M., & Wong, J. B. (2023). Screening for depression and suicide risk in adults: US Preventive Services Task Force recommendation statement. JAMA, 329(23), 2057–2067. https://doi.org/10.1001/jama.2023.9297