11/12 AI and Communication Technologies

communication technologies

any tool or system designed to facilitate the sharing of information or interaction between parties

forms of communication technologies in healthcare?

electronic health records (EHRs), telemedicine technologies (including chatbots), online pharmacies, telephones and fax machines

why are these technologies important?

• helps clinicans keep accurate patient records (can be transmitted to other providers in network + facilitate care)

• helps patients + clinicians coordinate care (online requests for med refills or scheduling appts)

• helps patients receive care at a distance (telehealth) (esp important for patients in rural/remote areas or patients who cannot physically mk it to an appt)

What are some key ethical issues in AI and Comm Tech? (Mehta 2014)

data security, privacy, clinician/patient relationship

ethical questions in data security

how do we keep electronic data safe? what do we owe patients whose data has been uncovered in a breach?

ethical questions in privacy

who has access to electronic data? what if someone other than the patient answers the phone when the physician calls?

ethical questions in clinician/patient relationship

do patients develop as much rapport with their clinician over a video call? to what extent can a clinician diagnose from afar (ex: diagnosing a skin condition over a televisit)

telemedicine (or telehealth)

the use of technology to provide care at a distance, such as video call (Mehta 2014)

telemedicine regarding patient differences

telehealth may not be a good fit for all patients, such as those uncomfortable with technology or who do not have access to internet, computers, or smartphones

telemedicine regarding evidence

how do telehealth interventions compare to in-person visits in terms of patient outcomes?

ethical issues in online peer support

T/F: Zoom support groups for mental health recovery became popular during the COVID-19 pandemic, presenting new ethical issues

true

what are new ethical issues that come with the covid pandemic?

• privacy (what if someone joins the calls from a public place, violating the confidentiality of everyone in the meeting?)

• DMing (how should the facilitators of support groups respond when someone receives an unwanted direct message from someone else in the meeting?)

what are the benefits of telemedicine/telehealth?

• allows patients to access their own records online, and to reach clinicians easily and quickly for help

• expands access to care for people who may not be able to get to a clinic or hospital in all cases

• can reduce costs associated with in-person appts (parking, bus fare, time away from work, etc)

• patients can take the call in a location that’s most convenient and comfortable to them

AI

technologies that allow computers to perform tasks that would typically require human intelligence

ex: a chatbot on a website that can answer basic questions about a clinic’s hours of operation

T/F: AI is a fast-growing health technology

True; more sophisticated AI can use complex algorithms to read diagnostic scans or detect trends in patients’ EHRs to identify possible diagnosis or indicate future risk

what are some ethical issues surrounding AI in healthcare?

informed consent, risk/harm, accountability

ethical issues in AI healthcare

informed consent

do patients understand the limits of AI technology? do they know they’re not communicating with a real person (chatbot), and do they know how their data will be used in the future to enhance the algorithm?

ethical issues in AI healthcare

risk/harm

are we putting patients’ health at risk, such as if an AI does not detect an illness that a clinician would have?

ethical issues in AI healthcare

accountability

if a patient is harmed due to the use of AI, can they sue for malpractice? who is responsible? oversight?

ethics and benefits of AI

improved accuracy, efficiency, increased knowledge

ethics and benefits of AI

improved accuracy

AI may be able to detect disease much earlier than clinicians can. improving possible treatment, and might be able to interpret finding with more reliability than a clinician

ethics and benefits of AI

efficiency

faster interpretation of diagnostic results, clinical trial data, and immediate sharing of info with patients

ethics and benefits of AI

increased knowledge

providing another perspective or source of data for clinicians and patients to make informed care decisions

summary of AI and comm tech ethics

• AI and communication tech can meaningfully enhance patient care + mk it more efficient, but should be used carefully due to potential ethical concerns

• data security, informed consent, risk/accountability, and the impact on the clinician/patient relationship r key issues

• how can we use these technologies in a way that advances patients’ interests and supports clinicians’ work?

discussion

an autonomy algorithm? (Lamanna and Byrne 2018)

would it be ethical to use AI to scour info abt a patient in EHRs and social media to inform healthcare decisions for them?

• many surrogates are overwhelmed by making decisions, or don’t actually know the incapacitated patient’s preferences

  • AI could predict preferences quickly and efficiently

    • can preferences truly be predicted by AI? are people’s social media posts truly reflective of their values?

    • would AI be influenced by existing human biases?

• no, there could be malware, nothing supercedes human judgment, social media is fake + posts r not truly reflective of a person’s values (ppl curate their identity, use sensationalization or satire etc. for entertainment etc.), outdated social media posts

• it def could, it could pick up on prexisting biases since that’s the data it’s trained on and could be prejudiced on how it approaches treatment

• needs oversight

• how much of your privacy are you willing to give?

"Telemedicine’s Potential Ethical Pitfalls" (Mehta, 2014)

• Core Argument: Telemedicine can improve healthcare access and efficiency, but it is only ethically acceptable if providers actively manage four potential pitfalls. These pitfalls are the erosion of the doctor-patient relationship, privacy threats, one-size-fits-all implementations, and assuming new technology is automatically effective.

• Supporting Evidence:

  • Tech Adoption Stats: 87% of adults use the internet, 58% own smartphones, and 72% of internet users search for health info online.

  • Policy & Literature: Cites a 2014 AMA policy stating telemedicine should supplement live visits for established patients, alongside systematic reviews showing mixed evidence on telemedicine's efficiency and outcomes.

• Unique Examples & Anecdotes:

  • End-of-Life vs. History: Contrasts using a digital questionnaire for routine medication history (appropriate) versus end-of-life discussions (inappropriate).

  • Automated Hovering: Mentions using remote sensors and automated reminders to engage patients during the "5,000 hours" they spend outside a doctor's office.

• What Stands Out:

  • Reframing Privacy: Argues data security is mostly an operational issue solvable with encryption, whereas the deeper ethical issue is patients not knowing who is viewing or answering their asynchronous messages.

  • Equity Paradox: Warns that while telemedicine expands geographic access, requiring computers or smartphones can worsen socioeconomic healthcare disparities.

"Should Artificial Intelligence Augment Medical Decision Making?" (Lamanna & Byrne, 2018)

• Core Argument: Machine learning can mine electronic health records (EHRs) and social media to create an "autonomy algorithm" that predicts incapacitated patients' treatment preferences more accurately than human surrogates. However, it must be used strictly as a decision aid rather than an automated replacement for human judgment.

• Supporting Evidence:

  • Incapacity & Surrogate Stats: Over one-third of elderly and psychiatric inpatients lack decision-making capacity, clinicians miss incapacity 42% of the time, and family surrogates guess patient wishes incorrectly about one-third of the time.

  • Predictive Data Studies: Cites studies showing computer models analyzing Facebook "likes" out-predict spouses on personality traits, and machine learning predicts cardiac surgery outcomes better than traditional regression.

• Unique Examples & Anecdotes:

  • Facebook "Likes" for End-of-Life Care: Suggests an algorithm could infer a patient prefers comfort-focused care if they previously "liked" the Death with Dignity National Center on Facebook.

  • Brain Cancer Speech Surgery: Illustrates training an algorithm on EHRs of brain cancer patients who could still speak to predict whether a mute patient would consent to risky speech-restoring surgery.

• What Stands Out:

  • The "Triple Burden": Frames incapacity as an ethical burden on healthcare systems, an emotional burden causing long-term trauma for surrogates, and an economic burden from unwanted overtreatment.

  • Bias Amplification: Warns that an algorithm might mistake doctor persuasion (e.g., urologists steering prostate cancer patients toward surgery) or low health literacy for genuine patient preference, reinforcing health inequalities.