Analyze text responses

M
Melissa
6 min read

The Text analysis module facilitates qualitative analysis of open-ended responses (text questions and comments) using artificial intelligence. It provides a quick summary of results and extracts themes, quote types, and sentiments to better understand the patient experience.

A Large Language Model (LLM) is used to extract sentiment from text by analyzing the tone, language and context of the words used.

⚠️
Model limitations

Although our goal is to provide you with relevant results, errors or omissions might happen in some cases:

  1. The context in which the comment was made is ambiguous
  1. Information is vague or inconsistent
  1. Subtle nuances haven’t been taken into account

In such cases, IA results might be inaccurate.

Text analysis attributes

Enable or disable text analysis on a question

The text analysis feature is enabled by default when creating a question. You can enable or disable it individually for text answer questions, comments and/or other choices.

  1. Access the Surveys list via the main navigation.
  1. Click on the survey you want to edit.
  1. Click on the Edit button.
  1. Click on the question you want to edit.
  1. Enable/disable the “Analyze sentiments found in the answers“ option using the switch (text answer questions and/or comments).
  1. Save the question.
💬
Answers and comments that are 4 letters or less are not analyzed, even if the switch is enabled.
⚠️
Answers and comments submitted when the text analysis is disabled will never be analyzed, even if the feature is enabled later on.

Overall Summary of the Analysis

Generate a summary

Save a lot of time and simplify your analysis by using our AI-powered Summary feature. In just a few clicks, get a summary of the last 5000 quotes attached.

  1. Access the list of Surveys via the main navigation.
  1. Click on the survey.
  1. Click on the Text analysis tab.
  1. If you wish, adjust your filters.
  1. Click on a Generate a summary.
  1. If you wish, click on the Copy-paste icon or on Export to export the summary in PDF format.
💬

The summary is interactive and allows you to access source quotes to validate the points raised by artificial intelligence.

  • Click on the Quote preview button to display the full list of segments related to this specific point.
  • Click on the View quote button to display the quote cited in the summary and access its context.

Summary content

The generated summary structures information into sections:

SectionDescription
StrengthsThe positive experiences that stand out most in the comments
Areas for improvementThe opportunities for improvement that emerge from the comments
TrendsThe topics or observations that emerge from the comments. These may be recurring patterns or isolated cases that merit attention.
Sentiment breakdownThe overall sentiment as well as the distribution of sentiments detected in the comments
Summary by extract typeThe summary of comments for each extract type

Detailed Analysis

After viewing the overview, you can deepen your analysis by exploring specific themes and individual comments.

Generate a summary by theme

Save a lot of time and simplify your analysis by using our AI-powered Summary feature. In just a few clicks, get a summary of the last 1000 quotes attached to a theme.

  1. Access the list of Surveys via the main navigation.
  1. Click on the survey.
  1. Click on the Text analysis tab.
  1. Click on the card Theme.
  1. Click on the theme you wish
  1. Click on Summarize.
  1. If you wish, use the Copy-paste icon.
  1. If you wish, click on Regenerate to get a different summary.

Filter quotes

  1. Access the list of Surveys via the main navigation.
  1. Click on the survey.
  1. Click on the Text analysis tab.
  1. Filter your quotes by a specific date or period of time.
  1. Click on Filters to see the answers to one or several specific questions, to exclude dropouts or to target a specific group of respondants.
💬
Find out more about Filters and cohorts in our dedicated article.

Export quotes

  1. Access the list of Surveys via the main navigation.
  1. Click on the survey.
  1. Click on the Text analysis tab.
  1. View the details of the text analysis by clicking on Themes, Comments, or Quotes.
  1. Click on Export.
  1. Choose your export parameters and click on Download.

Analysis indicators lexicon

Themes

Some quotes will also be automatically linked to one of the predefined themes listed below. If we cannot match a quote to a theme, it will be classified as “No theme”.

ThemeDefinition
Care and interventionsQuality and effectiveness of treatments, medical procedures, and care provided, including staff responsiveness and expertise with the patient.
Interpersonal skillsStaff interpersonal skills, including empathy, active listening, respect, and attentiveness to the patient's needs, emotions, and concerns.
WaitingWait times and delays associated with appointments, emergencies, and procedures, as well as the patient's perception of these delays.
InformationClarity, quality, and availability of information provided to the patient regarding their health condition, treatments, and procedures, including communication effectiveness.
SupportPsychological and emotional support offered to the patient, including the reassuring presence of staff and referral to appropriate support resources.
AccessEase of access to facilities, care, healthcare professionals, preventive services, and medications.
Comfort of the care environmentQuality of the layout, comfort, and quietness of facilities (rooms, common areas, treatment rooms), including the condition of the equipment.
PreferencesRespect for the individual choices, values, and personal, cultural, or religious beliefs of the patient in the delivery of care.
Pain managementAssessment, management, and effective relief of the patient's pain to optimize comfort and recovery.
CoordinationLevel of collaboration, communication, and organization among the various professionals and departments involved in the patient's care.
Continuity and transitionFluidity and follow-up during transitions between different phases of care, departments, or healthcare facilities.
Family and close onesQuality of integration, reception, and communication with family and loved ones in the care and decision-making process.
Quality of mealsQuality, taste, temperature, and nutritional suitability of the meals served to the patient.
Reception/admissionQuality, efficiency, and courtesy during the patient's arrival, reception, and admission process.
CleanlinessLevel of hygiene, sanitation, and maintenance of the facilities aimed at ensuring a healthy and safe environment.
AutonomySupport given to the patient to maintain their independence and actively participate in decisions regarding their health and treatments.
PrivacyRespect for the modesty, dignity, and privacy of the patient during care, examinations, and interactions.
LanguageAbility to offer services and communicate clearly in the patient's preferred language.
RightsRespect for the patient's fundamental rights, including free and informed consent, the right to information, and the protection of their integrity.
ConfidentialityProtection, discretion, and security surrounding the handling of the patient's personal information and medical record.

Quote Type

In addition to themes and sentiments, analyzed quotes are matched to a type that allows you to easily identify the nature of the comment.

TypeDefinitionExamples
Potential issueThe quote seems to indicate that the patient’s health and/or safety were endangered.• Mentions of medical or procedural mistakes • Perceived negligence that may cause harm • Mentions of acts of violence (physical, psychological or sexual) • Severe symptoms, significant emotional distress, suicidal thoughts or self-mutilation
PraiseThe quote seems to express gratitude for an individual, a team or a service.• A quote describing a positive interaction with a nurse or a doctor • Highlighting the quality of a specific service
SuggestionThe quote includes specific and achievable recommendations to improve patients’ experience. • Ideas for new policies or services • Suggestions for adjusting existing infrastructures or procedures • Suggestions for improving existing tools or technologies

Sentiments

Extracts from comments and open-ended responses are classified into 6 sentiment levels:

  • Very positive
  • Positive
  • Neutral
  • Negative
  • Very negative
  • Mixed

Did this answer your question?