Subhasha - Designing for language barriers in Indian Healthcare

Pre-thesis Project - Dec 2018 ||| Skills: User Research, UX Design, UI Design

This project aimed to understand the existing scenarios of healthcare communication with an intention of designing technology artifacts that could elevate the healthcare experience and strengthen the social threads.

This project was done as part of my Pre-Thesis project at college. To know more about the detailed thought process within every step of the process, please read my Medium articles about it here.

The overlying theme of our project was ‘Weaving the Threads’. It looked at tightening the social fabric by strengthening bonds between the people in this fabric. There were so many possibilities within this brief.. so why did I choose the topic of language barriers in healthcare?

  • Language is an integral part of human interactions having a lasting impact on the social fabric. Yet with its evolution barriers have emerged in varying contexts.
  • The Language Action Perspective (a theory introduced by Flores and Ludlow) says that people are linguistic beings and use language to perform actions. Thus the barrier to language often acts as an obstruction to this completion of action.
  • Through secondary research it was seen that healthcare is a field where in the repercussions of the language barrier were very serious.
  • Research work on this topic was mainly done in Western Countries and hence there is a need for relevant research in the Indian context.

Exploratory Research — Understanding the Context

After a literature review to understand work that has been carried out in the context of healthcare communication, I felt it was necessary to carry out a primary research study to understand specifically the Indian context.The primary research study was done through Contextual Enquiry and Telephonic Interviews. You can find the questions planned for the interviews here.

Consolidation and Analysis of Data

In order to make sense of the vast amount of data that was collected as part of the research study, I needed to first consolidate it and then map my insights. I chose to do this using the below format for every user and then find the patterns across users. Since I was doing this process individually, I chose to do explore the possibility of consolidating it using digital tools rather than usual post its and large charts.

Research findings

There were many different insights that came out of my research. In order to make it clearer for me to take this ahead and to effectively communicate it to anyone else interested in this study, I consolidated it into 3 main categories. Each of the findings were driven by insights dervived by seeing patterns across all the users.

  • Situations where the problem of a language barrier exists
  • The implications of the language barrier
  • The ways in which this problem is being handled at the moment

Situations where the problem of a language barrier exists:

Clinical settings- Govt and private OPDs, clinics and hospitals: India being a multilingual country with high internal migration, there is a mixing of cultures with variety of languages spoken in an area. Each of the findings were driven by insights dervived by seeing patterns across all the users.

Organisations of doctors who travel to remote areas and crisis situations to provide medical aid This includes voluntary organisations (Eg. Doctors for You) where in the doctors travel to areas that they do not know the language of.

Community healthcare programs in rural areas: Public health programs are carried out by non local community health personnel in villages to inform the people about risk factors for non communicable diseases.

Implications of the barrier:

Increased time needed: In cases where there is a language barrier between the doctor and patient, the time involved in completing the actions of diagnosis and treatment is much higher than usual. Increased time was also mentioned in the field of community health.

“If we see that there is a language or communication barrier we ask the same questions in many different ways” — U01 ( A practising doctor )

Decreased transparency: TThough there is a clear need for the details of a diagnosis to be completely clear to the patient at the end of the interaction, language or communication barriers can cause less information to be disclosed, something that is far from what is the desired action. In this way, the moral right of a patient to know about their own illnesses is compromised.

“Kabhi hindi mein likha hua nahi mila.. Unhe english aati hai so likh dete hai.. Par mujhe bhi toh dekhna hai na kya bimari hui hai” -U12 (Patient)

Possibility of loss of information: When there is a communication barrier, it is seen that even if the actors manage to communicate, it is sometimes not complete or effective. Other than language barriers, it was seen that medical terminology also forms a sort of communication barrier between doctor/medical personnel and lay man. This results in the patient only understanding some part of the intended message.

“More than understanding the patient’s problem, it becomes difficult to explain the treatment, especially in cases where surgery or some other follow up is required.” — U03 ( Practising doctor )

An interesting aspect about this loss of information was looking at the concept of interpreters. Although the use of formal trained interpreters is almost an unheard phenomenon in India, many of the doctors and patients mentioned that often informal interpreters are used to convey the message. This again adds another layer of interpretation and possibility for loss or change of message. Details on types of interpreters are explained further in this article.

“Knowingly or unknowingly the translator maybe causing misinterpretation. This can cause a lot of problems because it can impact treatment and the patients think the doctor wasn’t good.”- U08

Impact on trust and relationship building:The previous implications all looked at the functional, quantifiable measures of understanding but all of them finally lead upto a larger social impact. The trust and comfort levels between the doctor and patient are surely affected by a language barrier.

Present ways of handling the barrier:

Models and images It was found that specific models are made and provided by medical representatives of various companies. The types of models or charts vary from field to field with a seemingly larger use in orthopaedics. However, this wasn’t a pattern across all users as some doctors also stated that they don’t usually use these tools. Patients too didn’t seem to recall this as an important part of their communication experience.

Informal interpreters:

Gestures and actions: Drawing from the interviews conducted, using actions is found to be the most common first reaction to dealing with verbal communication barriers. Patients mentioned ‘pointing to the area of pain’ as a way to explain the symptoms while doctors used common gestures to ask questions or convey treatment information.

Adaptation of written information like prescriptions: This is an area of the findings that had mixed responses from patients and doctors. While many of the doctors talked about writing instructions on the prescription in local languages like Hindi, the interviewed patients said they have never received a prescription in a language other than English.

At the same time, it is important to note that U09 mentioned a study he had carried out, which found that 80% of the prescriptions written by doctors in India do not follow the guidelines as decided by the medical council and taught to them during MBBS.

Biggest learning from the research phase

“You will start with one problem and still find an ocean of problems along with it. And it’s ok if you can’t solve them all, as long as you are aware of their existence! Try to solve for one small thing.. but do it with all your heart!”

Ideation Phase:

From my findings from the research phase, I had 3 main ideas that looked at different scenarios and problems. In order to depict each one of them, I made storyboards so that I could take this back to the users and gain concept level feedback at the initial stage itself. By communicating my ideas through these artifacts (storyboards) it was easier to have a more engaging conversation and these sessions served as further research on the context as well.

Idea 1: Technology for access to interpreters & terms

  • Technology as a tool to connect humans and create collaboration
  • Access to interpretation services on demand and based on severity of need
  • Building upon existing knowledge base of professionals

Idea 2: AR in community health programs

  • Using existing posters to create more interactive communication
  • Dynamic visual & multi-lingual content act as aid to the conversations
  • Common representations for both actors (community health personnel and villager) thus improving knowledge transfer

Idea 3: Communicative Prescription Making Process

  • Providing resources to make this a more interactive process that would help enhance this conversation of explanation of diagnosis and treatment.
  • Prescription as an artifact that helps the communication inside and outside the clinic
  • Mix of visual and written content that also facilitates better verbal communication between actors
  • Standardised record, more transparency and easier for patient to understand

Concept evaluation

As a means of gaining early stage feedback on the ideas being explored, the storyboards were taken to doctors and patients and a means to have further conversations about the existing situation and such possibilities. The concept evaluation sessions were carried out with 6 doctors and 2 patients.

*Unfortunately due to logistical difficulties, I was not able to do a session with the community health personnel to gain feedback on the idea for that space. (Idea 2)

These interactions included:

  • Explaining the storyboard
  • What could work and what won’t?
  • What other factors should be considered?
  • Conversations about the problem space that arise out of engaging with the concepts
  • Interacting with artifact and talking about it

Drawing and explaining

As a means of gaining early stage feedback on the ideas being explored, the storyboards were taken to doctors and patients and a means to have further conversations about the existing situation and such possibilities. The concept evaluation sessions were carried out with 6 doctors and 2 patients.

Role Play experiments

At the end of this phase, for each of the ideas I had some feedback which helped me decide what idea or parts of an idea I can take forward. I wanted to explore the idea of visual aids as a means of communication a little more in detail before I could go ahead with the concept. For me to evaluate the actual usage I decided to conduct some role play experiments.

Pilot Session
To evaluate how effective visual means of communication is to explain diagnosis and treatment,
(1) when there is a complete language barrier between doctor and patient
(2) language is common but some communication barriers exist due to patient background

"People create their version of understanding by stitching together the few terms they understand in an otherwise alien conversation"

Modification to experiment: Visual + Verbal Aids?
Since visual aid did not seem to work independently, the possibility of the doctor having access to a specific list of terms relevant to the diagnosis along with the visual aids was explored.

Experiment Session: Visual + Verbal Aids
These experiments were aimed to evaluate 2 main factors:

(1) Impact of availability of terms in patients language along side sketching tools while talking to the patient
(2) Influence of sketching from scratch vs sketching using base images on the communication process Details of actors

  • Dr Manjari is trying to understand symptoms, explain diagnosis and treatment to a patient who doesn’t share a common language with her.
  • 4 patients who were all part of the housekeeping staff at Srishti Institute were recruited for these sessions.
  • Sujatha, the nurse, helped act as the interpreter to recruit and communicate to me how much the patient actually understood

Main learnings

These experiments showed that the access to visual and verbal resources did help the doctor and patient understanding was increased. However, there were some more nuanced observations that led to insights for the concept to be developed further.

  • There are 3 main types of terms that are useful:
    a) General terms common to all diagnosis — left, right, pain, how many
    b) Organ and system names — kidney, urethral tube, etc
    c) Terms specific to that diagnosis- Stone, water, flushing out, etc
  • Gestures, sketching and verbal communication work together to form any kind of conversation and cannot be looked at independently
  • Relevance to ongoing conversation is important for the resources to be totally useful
  • Explaining pronunciation is an integral factors that needs to be considered

Final concept: Subhasha

Subhasha aims to provide local colloquial language and image based intelligent assistance for doctors and patients. It hopes to enhance verbal communication despite the language barrier between a doctor and patient during the diagnosis and treatment phase for diseases of moderate severity.

Main features

These experiments showed that the access to visual and verbal resources did help the doctor and patient understanding was increased. However, there were some more nuanced observations that led to insights for the concept to be developed further.

  • Locally crowd sourced terms to make sure they are colloquial and chances of patients understanding them is higher
  • Background algorithms with reinforced learning for continuous improvements with prolonged usage
  • Visual resources and terms relevant to the stage of conversation i.e. pre diagnosis questions and post diagnosis explanation of the problem and treatment
  • Glanceable interface with limited interactions unless the user wants to, so that the conversation is not disturbed
  • Dynamism created by various factors including on going conversation (voice input), interactions with the screen (taps and sketching) and previous records of that patients
  • Share with the patient to improve patient understanding outside the clinic as well.

Actor Algorithm Information Architecture

The actor algorithm interaction architecture was made to clearly define the interactions between the actors using the system and the system back-end via the medium of the application interface. This has been presented in the form of the four main layers:

  • Interaction and conversation between the actors (Doctor and patient)
  • User facing interface
  • Immediate backend system
  • Constant learning core

By laying out the detailed connections between these layers at every step, the entire journey of usage has been elaborated. Creating this artifact helped me iterate on the issues that may arise, the further possibilities and gaps in the system.

I know that it’s impossible to read… But then detailing out interactions means something big coming out! So just a zoomed snapshot of one of the parts so you can get the idea of what it was like!

Application Task Flow

While designing the screens, there were various considerations to be thought of so that the UI helped achieve the goals of the project.

UI design

The interactions within the entire system were detailed out but from an interface point of view, I needed to define the steps and actions the user could perform on the app.

  • Flexibility of interface based on choices made by the user such as only terms, hiding details of usage, change of diagnosis, etc
  • Relevance of dynamic content and clarity for user to use it effectively.
  • User control to prevent system from listening if they don’t feel comfortable
  • Allowing for effective use whether first time or repeat patient
  • Glanceability of the interface so that the doctor can use aid without disturbing the physical ongoing conversation

Material design

The entire UI design followed the principles of material design including spacing, layouts, elements, colour and typography. It was a great opportunity for me to understand the material design system more deeply and find ways to maintain consistency while fulfilling the specific needs of my application.

Iterating on the UI

Various iterations for the different screens of the application were explored first through wireframing and then digitally. Below are some of the earlier explorations

Prototype version 1

In order to be able to simulate the flow and interactions of the user with the interface in the context, an interactive prototype was made using Adobe XD. Most of the interactions were simulated within the prototype. For live data based dynamism, representative animations were made based on dummy content.

The prototype was made on the assumption that the doctor is a gynaecologist who speaks Hindi and English and the patient speaks only Kannada. The diagnosis worked on was PCOS (Poly Cystic Ovarian Syndrome).

User Testing — Cognitive Evaluation (Adapted)

Tasks were given to users and they interacted with the prototype to try and complete those tasks. They were imagining the scenario because it wasn’t an actual working prototype to test with patients but since it was a usability testing of the interface I could still gather insights for improvements in the UI. The misunderstandings and gaps between their conceptual model and the proposed conceptual model match were evaluated.

Prototype 2- Changed screens

I hope to use this prototype to carry out more longitudinal studies to evaluate its usage within the context. At the same time I have also looked at the technology development bits to look at understanding the working of the back end system.