This article is more than 5 years old. Information may no longer be current.
Artificial intelligence increasing in community oncology, but cannot replace physicians
Click Here to Manage Email Alerts
ORLANDO — The use of artificial intelligence in community oncology practices has the potential to grow rapidly and provide more assistance to oncologists and hematologists in the coming years, according to presenters at Community Oncology Alliance Annual Conference.
But artificial intelligence (AI) will not take the place of oncologists, and presenters warned of relying too heavily on AI when caring for patients in the community practice setting.
“We have to find a way to be very agile and adaptable,” Aaron Lyss, MBA, director of strategy and business development at Tennessee Oncology, said during a presentation. “One of the things you will see in the community practice setting that’s different than the academic setting, to use a baseball metaphor, is that we will not be swinging for the fences and missing. We are going to be [more careful] and hit doubles and singles.”
The ability of AI to assist physicians was demonstrated in a recent 8-month pilot study involving Northwest Medical Specialties and Jvion’s AI machine. The goal of the study was to reduce costs and provide better quality care to patients at high risk for mortality by using AI to assess which patients were better suited for end-of-life care instead of continued treatment and looked at ways to improve end-of-life care to make patients more comfortable.
Certain “vectors” the study assessed included risk for 30-day mortality, risk for severe or moderate pain in the next 30 days, risk for depression diagnosis within 6 months, risk for deterioration of activities of daily living (ADL) levels within 6 months, and risk for avoidable and multiple admissions.
Interventions recommended by AI included re-evaluating treatment plans; encouraging advance care planning; preparing patients, families and caregivers; focusing on symptom management; and considering mobilizing community support for the patient.
Results showed a 225% increase in hospice referrals (0.1 per 1,000 patients monthly to 0.5 per 1,000 patients monthly) and a 35.3% increase in palliative care referrals and supportive care consults (8.4 per 1,000 patients monthly to 11.3 per 1,000 patients monthly).
The intervention led to a 30% overall reduction in deterioration of ADL levels within 6 months and a 33% reduction in moderate or severe pain within 30 days. It also resulted in a 22% increase in depression diagnoses within 6 months.
“It picks up stuff that you don’t know,” Sibel Blau, MD, medical director at Northwest Medical Specialties, said during the presentation. “It’s something that you can use [as part] of your practice.”
Another pilot study, presented by Ray Page, DO, PhD, president and director of research at the Center for Cancer & Blood Disorders, had a similar setup and found that hospice referrals increased 113.3% (0.01 per 1,000 patients monthly to 0.03 per 1,000 patients monthly) and palliative care referrals increased 218.8% (0.03 per 1,000 patients to 0.08 per 1,000 patients every month).
It also demonstrated a 17% reduction in 6-month deterioration of ADL levels, a 28% reduction in moderate and severe pain over 30 days, and a 33% increase in depression diagnoses over 6 months.
“I think [AI] here in 2019 [should be] used as a nudge to provoke our thinking about patients,” Page said during the presentation. “It does not replace physicians or our judgement. Sometimes it will change your thinking, and sometimes it won’t.” – by John DeRosier
Reference:
Blau S, et al. Artificial intelligence applications in community oncology. Presented at: Community Oncology Alliance Annual Conference; April 4-5, 2019; Orlando.
Disclosures: HemOnc Today could not confirm the authors’ relevant financial disclosures at the time of reporting.