Tools
This is the oldest type of AI systems that mimic or support the decision-making abilities of human experts. Uses knowledge base, rules and criteria.
What is expert systems? or What is clinical decision support?
This is the “raw material” AI needs to learn—everything from vital signs to notes to images.
What is data?
This AI tool scans patient data to warn clinicians early when a patient may be developing sepsis.
What is a sepsis predictive model?
Identifies and collects data from multiple places and then notifies the clinical team, including nurses, so we can intervene sooner.
This ethical issue occurs when an AI tool performs worse for certain patient groups because the data used to train it was not diverse.
What is bias?
This type of data is organized in a fixed format with labels, like vital signs, lab values, or medication lists.
What is structured or descrete data?
Descrete data is easily extracted and placed into tables for analysis.
This AI tool enables machines to interpret and process visiual information from the world.
What is computer vision?
This is a step where AI looks for patterns in data so it can make predictions—similar to how nurses learn from experience.
This type of AI listens during patient care and automatically drafts documentation for the nurse.
What is ambient listening?
Very big for providers currently, making it's way to nurses.
This principle means that nurses should understand, at a basic level, how an AI tool reaches its recommendations before relying on it.
What is transparency?
This type of data includes free text nursing notes, scanned documents or images.
What is unstructured data?
Unstructred data is not easily extracted and will need to be processed before it's able to be utilized.
A subset of AI where systems learn patterns from data to make predictions or decisions with out explicitly being programmed?
What is machine learning?
This part of AI does the actual “thinking”—it takes in new information and produces an output, like a prediction.
What is a model?
The model, uses what it learned during training to make decisions or predictions.
Computer vision AI tools help nurses by analyzing photos of wounds to estimate size, depth, and healing progress.
What is AI wound imaging or wound measurement?
Yes! This exsists and is coming to a bedside near you!
This ethical responsibility requires nurses to confirm AI suggestions with clinical judgment rather than accepting them automatically.
What is accountabilitiy?
This phrase means that poor-quality data leads to poor-quality AI output- no matter how good the model is.
What is garbage in, garbage out?
This subset of AI can create new content such as text, images or summaries based on patterns it learned from large amounts of data.
What is generative AI?
Another level after machine learning. Produces new outputs rather than just learns, predict, and/or analyzes.
This is the instruction set or "recipe" that guides how an AI model processes data and makes decisions.
What is an algorithm?
This how models are told what to do with the data. Can include complex criteria.
This AI tool creates patient assignments, predicts census, and staffing needs.
What is an AI staffing prediction tool? or What is an AI assignment tool?
When patients receive information from an AI chatbot, the organization must ensure the content is accurate and appropriate. This ethical principle requires humans—not the AI—to remain responsible for the final information shared with patients.
What is human oversight?
This is why validation is so important as well as accountability. If the chatbot gives advice and the patient follows it. Is that the organization's responsibility or the chatbot tool creater/owner? Lots to ponder!
This is a common source of bias in AI models when the data does not represent all patients or groups equally.
What is non-representative data? of What is sampling bias?
What is Natural Language Processing?
Examples include speech recognition and language translation.
This checks how well an AI tool performs and helps ensure it works accurately across different patient groups.
What is model validation? or What is model evaluation?
Tests that the model is doing what we are asking it to do while ensuring it is reliable, fair and safe.
This AI tool helps reduce nurse workload by answering basic patient questions, providing education, prompting patient follow-up, collecting intake information, or helping with scheduling.
What is a chatbot? or What is conversational AI?
This concept reminds us that even helpful AI tools can unintentionally reinforce health disparities if not monitored and validated across populations.
What is fairness? or What is equity?
This essential concept descrives how well the data reflects the real-world clinical situation, such as accurate documentation or correct patient labels.
What is data quality?
High quality data is complete, accurate, and consistent which leads to less errors by AI.