Using Personal Social Networks to Tailor News to Family and Friends

Wendy Moncur, Universities of Aberdeen & Dundee, Scotland, UK. wmoncur@abdn.ac.uk

Problem and Motivation

When a newborn baby is admitted to a Neonatal Intensive Care Unit (NICU), friends and family members are likely to be concerned. They want to know how the baby and the parents are doing, and what assistance they can give [1]. It can be difficult for them to get this information, yet it is essential if friends and family are to give practical and emotional support to the parents. This support helps the parents to cope with stress [2].
The information is difficult to obtain for a number of reasons. Hospitals are unable to give out news about patients to friends and family. The parents will not be contactable if they are with their baby, as use of mobile phones is not allowed in NICU in the UK. Even when parents do go away from their baby and the NICU, they may struggle to respond to multiple messages asking for information. They are likely to be exhausted from childbirth, and anxious about their sick infant. The mother may herself be very unwell.
I am developing a computer system, BabyTalk-Clan, which provides information to family members and friends in these circumstances. The term Clan covers the people who we know individually and have a personal relationship with [3]. A key challenge in BabyTalk-Clan lies in deciding what information should be given to each clan member. In this paper, I suggest that a “ first pass ” decision can be inferred from the parents’ personal social networks. Parents may subsequently refine the report content or recipients if they wish. A simple, intuitive tool is described that captures this social network and infers from it the amount of information to disseminate to individual clan members. For simplicity within the paper, the existence of a mother and a father will be assumed, although it is recognised that this does not reflect the true diversity of family compositions.

Background & Related Work

Research Context

BabyTalk-Clan is one of a group of systems being developed in the BabyTalk project at the University of Aberdeen, in collaboration with the Royal Infirmary of Edinburgh and the University of Edinburgh. The systems will provide information about babies in NICU. The architecture of the system uses a combination of Artificial Intelligence (AI), Signal Analysis and Medical Reasoning techniques. The information is derived from continuous sensor readings (such as heart rate), and discrete event records (including medication administered, equipment settings and laboratory results) [4]. These are used to generate appropriate textual summaries for diverse users - doctors, nurses, parents, friends and family - using Natural Language Generation, a branch of AI .


Figure 1: Sample extract from Parent Report, for ‘test’ baby George, with hyperlinks to explanatory text underlined.

George has been changed from nasal CPAP to ventilation. He has been in oxygen between 25 and 40%. George has been given a dose of replacement surfactant. He has been started on morphine to keep him comfortable. He has been started on vecuronium to keep him from moving and fighting the ventilator He has been started on dopamine to help his blood pressure.


Figure 2: Sample extract from Baby Diary for ’test’ baby George.

Dear Mummy and Daddy,
Thank you for my cuddly dog – I’ve been having big, long chats with him and he's been in my nest with me. He even let me suck on his ear! Nurse Sophie gave me some of Mummy's milk from a cup but not enough so I shouted until she gave me some more. Now I get enough milk to fill my tummy right up so I can sleep for longer and I can give my doggie's ear a break!!

At the moment, parents can get updates from medical staff caring for their baby, and from a computer-based information system which generates two very different kinds of reports for parents [5]. The "Parent Report" (Figure 1) is generated automatically, largely from clinical data. It is serious in tone, accurate, and has hyperlinks to definitions of technical terms used. The non-technical "Baby Diary" (Figure 2) is written by nurses, tends to report on good news only, and also contains photos of the baby. Access to these reports is via a secure web-link. In the absence of an information system designed specifically for clan members, some parents have shared their login and password information for the Baby Diary, particularly when friends and family were geographically distant. This sharing is legitimate: in the UK, patients – or their parents, if the patient is under 16 years old – have the legal right to disseminate their medical data as they wish. When talking about sharing information, parents have said that it would be useful if updates about the baby could be tailored for different individuals, rather than having to choose between the Parent Report and the Baby Diary. For example, they would communicate bad news and details of medical interventions to selected individuals, but not to all.

Personal Social Networks

Personal social networks consist of the people that we know. They have a characteristic, hierarchical organisation [3, 6]. Membership of a specific layer of the hierarchy is influenced by emotional closeness and contact frequency. The innermost layer, the Support Clique, consists of 3-5 stalwart individuals that we feel emotionally closest to. They provide us with advice, support or help at times of acute distress. The next layer, the Sympathy Group, typically contains 12-20 individuals who constitute the foremost friends and relatives. The Band consists of 30-50 people that we commonly associate with. It is an unstable grouping, where the changing membership is drawn from the Clan, itself made up of the people who we know individually and have a personal relationship with (around 150 people). Note that each of the membership values is inclusive of the layers within them. This hierarchical organisation is common regardless of gender, age, personality and culture, although these factors do affect the size and membership of each layer. For example, men tend to have more males in their networks, women tend to have more females.
Within a social network, individuals give and receive social support in the form of emotional, material and informational aid that is tailored to the individual recipient. In this way, network members build up ‘network capital’. Despite the structural consistency of personal social networks, this network capital varies by gender, socio-economic status and age [7]. Women tend to give more emotional support than men. People in low socio-economic groups may generate larger amounts of network capital because they have little financial capital, and therefore need to trade favours to acquire practical services. Members of a social network who share a similar situation may also build greater network capital. For example, parents of young children may discuss parenting experiences together (emotional support) and baby-sit each others’ children (practical aid).
In the event of an adverse health incident, social support provided by friends and family has beneficial effects [2]. Social support reduces stress. It can promote a more positive interpretation of events, and encourage coping strategies. Those who perceive themselves as receiving a high degree of emotional support during adverse life events even live longer. Emotional support can remind the recipient to care for themselves despite the problem being faced, for example by eating and sleeping properly. For mothers of infants, social support improves mothers’ sense of life satisfaction, and gives them a more positive attitude to the experience of being a parent [8].
Despite these established benefits of social support, there is minimal research into using computers to actively stimulate provision of this support amongst a patient’s personal social network during an adverse health event. In contrast, much effort has been invested in the creation of on-line patient support groups, despite arguments that manufactured patient support groups deliver less benefits to the patient than those offered by a patient’s existing personal social network [2].

Uniqueness of the Approach

Rather than look at how to improve what was communicated already by the BabyLink reports, an inductive approach was adopted to capture parents’ and clan members’ information requirements. A technique was adapted from one used by childbirth educators when helping expectant parents to analyse potential sources of emotional and practical support. The technique uses the mother’s and father’s shared personal social network as a reflective device [9]. Highly visual, the technique does not rely on language or literacy skill, nor is it culture-dependent. It was adapted into a prototype tool for NICU parents.
The prototype tool used familiar, everyday materials: a mat marked with concentric circles and a selection of buttons and sequins. Each parent was asked to choose a button to represent their baby, and place this at the centre of the set of concentric circles on the mat. They were then asked to select further buttons to represent themselves, their partner, friends, family and colleagues, and to place these at a distance from the baby to represent how close these people were at the time of the baby’s stay in NICU. The term ‘close’ was not defined unless participants requested clarification: we preferred people to interpret it for themselves. If they did ask for clarification, ‘close’ was defined as how involved people were in the situation, either emotionally or at a practical level. After creation of their map, each subject was interviewed individually.

Results

The tool was piloted in a small study with an opportunity sample of parents whose babies had been in NICU, consisting of 2 men and 5 women. This included one couple. Five parents were educated to university degree level, two to high school level. All were employed. Parent’s age at time of the baby’s birth ranged from 18 to 45. Of the babies discussed, there were 5 singletons and one set of twins. Earliest gestational age at birth was 32 weeks. One of the babies had long term health problems, six were normal at two years after birth. Participants were visited at home, or invited to the researcher’s home, to create a relaxed environment for discussion, and to facilitate easy childcare for the participants.

Effect of gender

Whilst the number of subjects was too small to draw firm conclusions, initial results suggest that the complexity of social networks, and the nature of communication by parents when their baby is in NICU, may be affected by gender. All of the women interviewed perceived their own mothers to be very close: men did not. The five female subjects carefully created complex maps of individuals. They took great care in representing individuals with appropriate buttons, for example using a tiny, exquisite bead for the baby and a chunky brown button for an elderly relative (Figure 3). The two male subjects took a more ‘broad-brush’ approach, mapping groups of people as a single node, using whatever buttons came to hand. Moreover, distance on the map between individuals was measured more carefully by women than by men. These results are consistent with findings that women tend to be more aware of the subtleties of interpersonal relationships than men [2].

Personal social network for mother

Women appear to hold a more immediately accessible mental map of their social network than men. Prompting for greater detail in maps led women to add in ‘official’ figures that they perceived as supporters – for example, clergy. They did not add extra friends or relatives. In contrast, men added some of the nuance previously missing from their maps, identifying male friends or relations who were closer than the rest of their network. This suggests that men may need more prompts in an interface for inputting their social network than women do.

Information broking

Mothers provided some information to clan members themselves. They wrote 'thank-you' cards, incorporating updates on the baby, and spoke to hospital visitors. Apart from this, mothers relied on one or two nominated ‘information brokers’ to disseminate information for them. These brokers were the closest to the mother and baby on the mother’s map – usually her partner and the baby’s maternal grandmother. The broker(s) acted as a contact point for clan members, protecting the mother from a barrage of solicitous enquiries. One husband carried his wife’s mobile phone as well as his own, as she was hospitalised and therefore unable to use it herself. Once information brokers had passed news to some clan members, “the message filtered out” from these members to other clan members further out on the map, in a ripple effect.
The degree of satisfaction in information brokers varied. While women tend to communicate about thoughts and feelings, men tend to focus on the pragmatic [7]. Fathers could therefore disappoint in the role of information broker, communicating different information to that which mothers thought important. Describing her husband’s ability to communicate detailed updates to supporters, one mother said “You know how it is with men… they don’t tell you anything…”. In contrast, another mother particularly valued the communications of the maternal grandmother, who emailed photographs to clan members: “three months later… they still had the picture on the notice board that my mum emailed of the twins”.

Factors that affect information dissemination

Information dissemination was affected by intimacy. Parents gave different information to different people. It was only partners and the maternal grandmother who were given detailed information when there were serious health worries: others were not told. Close clan members asked detailed questions about the baby and mother, and were identified as providing most emotional support to the parents. One hospitalised, bed-bound mother sought emotional support by showing her Caesarean-section wound to a close female friend, and discussing how she felt about it. Understandably, she did not do this with all her visitors.
Dissemination was also affected by demand for news, and by time. Those further out in parents’ networks either “weren’t as interested” in the baby, and did not want as much detail, or “tended to panic” and were deliberately given less information. When brokers were short of time, they gave less information to these people. However, when appropriate information was not given out to clan members at all, there could be adverse effects. For example, mothers in hospital were deluged with visitors, which they found unwelcome and exhausting.

Discussion

It is essential that automated updates generated for clan members do not discourage appropriate human contact. The objective of the tool discussed is to provide members of the social network with information so that they do give valuable social support. Parents gained a sense of support from people’s interest in their baby’s wellbeing. Conversely, when clan members failed to demand information, some parents felt unsupported and distressed.
Results of the prototyping exercise and interviews showed that the layered bands of social networks can be used as a model for controlling information dissemination to clan members (Table 1). The information that clan members want has previously been established [1]. This is used here as suggested input to the model (Table 1). The amount of information that an individual would receive is cumulative: for example, a member of the Sympathy Group would receive all of the information for Clan, Band and Sympathy Group members.


Table 1: Cumulative information dissemination by hierarchy layer
Hierarchy Layer Information disseminated
Clan Brief alert to say that the baby has been admitted to NICU.
Baby’s date of birth, gender, name.
Is it ok to visit, & visiting times if appropriate.
Band Is Mum in hospital.
Which hospital is baby in.
Sympathy
Group		 What’s wrong with the baby (high level explanation).
Change in baby’s condition.
What practical help do parents need.
How to contact parents.
Does the baby need help with breathing.
Does the baby need surgery.
Support
Clique		 How are parents coping emotionally.
Is it an inherited condition.
What’s wrong with the baby (detailed explanation).

Work is in progress with parents of current NICU babies to test out the model further, prior to implementing it. These outstanding questions will be pursued:

Contributions

This paper explores how the dynamics of the social network affect the design and dissemination of information. It considers an approach that could have a benign effect on the lives of everyday people experiencing a stressful life event, who would benefit from social support. The life event need not be limited to having a sick infant in NICU. The approach can be generalised. For example, for patients with chronic conditions like cancer, practical and emotional support is essential over an extended period.
The prototyped tool is innovative in its simplicity. It makes no assumptions about the people that constitute an individual’s personal social network, and can be used with diverse cultures and family structures. Once the system is implemented as a computer-based application, it is intended that users download their contacts from their mobile phone and then arrange them on their personal social network map, using a graphical user interface version of the prototyped tool. The tool provides a potential extension of the functionality offered by popular social networking sites such as Friendster.

Acknowledgements

Thanks are due to the parents who took part in this research. I would also like to thank Dr.Ehud Reiter, Dr.Judith Masthoff, Professor Ian Ricketts and Dr Yvonne Freer for their valuable guidance. This research was supported by the UK Engineering and Physical Sciences Research Council, under grant EP/D049520/1 and a doctoral training award.

References