S3.E6. 解讀隱藏的訊息 Deciphering hidden signs

國際學術研究團隊與百靈佳殷格翰合作，使用人工智慧和先進的電腦程式，一起分析語言模式，進一步評估青少年是否有罹患思覺失調症等精神疾病的風險。

Sidarta Ribeiro做了詞彙統計，將句子以各自的單詞劃分，並將內容與文字重組成新的內容。一位讀者試圖在Ribeiro的螢幕上解讀這個混亂的單詞謎語，他可能會認為這毫無意義並且很快就會放棄。「這正是我們研究的關鍵部分」，Ribeiro解釋道。「比起內容，我們對詞語相互作用的結構更感興趣。」

單字，音節，句子：對於研究者來說，這些都是用來提供智慧軟體的主要數據。這語音檢測工具的程式，能識別混淆人類耳朵和眼睛的單字、聲音或是音節中的模式和邏輯。

巴西神經科學家與橫跨多領域的團隊評估的數據，來自心理學家根據特定問題與試驗，和參與者進行的對話錄音。「舉例來說，醫生要求參與者討論他們最近做的一個夢」，負責百靈佳殷格翰幾個中樞神經系統（CNS）臨床項目的Michael Sand博士解釋說。參與者還被要求要描述具有情感背景的某些圖像。

Ribeiro研究小組的成員使用專門設計的軟體，對這些錄音對話進行評估。「對思覺失調症患者的研究說明，這種疾病的跡象反映在他們的言語模式中」，Sand表示。語調的變化和言語複雜程度降低的狀況，都可能是疾病的早期症狀。

百靈佳殷格翰和該研究小組目前計劃利用這些發現，來幫助並及早診斷相關疾病。 Ribeiro說明：「容易患有這種疾病的人當中，思覺失調症的發病通常不會比青春期發生得早。」他補充說道：「思覺失調症是一種隨著時間變化，進而變得更糟的疾病」。但如果能及早被診斷出來並得到適當的治療，惡化的過程可能會減緩甚至停止。

Ribeiro樂觀的認為，醫生將在不久後能藉由他在研究中使用的智慧軟體，對此疾病的高風險族群進行評估。雖然過去需要花費數天或數週的時間來分析語言中的模式，但現在只需要按幾個按鍵、花費幾秒鐘的時間就能完成。「由於數位化，使我們的研究變得更快」，Ribeiro說。「希望許多患有此疾病、高風險的人能因此受益。」

(本文單純報導百靈佳殷格翰在國外的科技創新與應用，目前台灣並沒有該應用的導入，同時呼籲：疾病的預防與治療都應與專業醫療人員配合，尊重專業意見，不宜自行判斷。)

Using artificial intelligence and advanced computational methods, an international academic research team, in collaboration with Boehringer Ingelheim, analyses speech patterns in order to assess whether adolescents are at risk of developing mental illnesses, such as schizophrenia.

Sidarta Ribeiro counts words, chops up sentences into their individual elements and shuffles texts into new texts. A reader seeking to decipher the chaotic-looking results of this word puzzle on Ribeiro’s screen would probably consider this pointless and quickly give up. “This is a key part of our project,” explains Ribeiro. “Here, we are interested less in the contents of the texts than in the structure of the interplay of words.”

Words, syllables, sentences: for the researcher these are primarily data which he can use to feed the intelligent software. The algorithms of this speech detection tool identify patterns and logical contexts in words, sounds and syllables that elude the human ear and eye.

The data that the Brazilian neuroscientist evaluates with a multidisciplinary team come from recordings of conversations which psychologists have conducted with trial participants on the basis of a specific system of questions. “Doctors ask the participants to discuss their most recent dreams, for example,” explains Dr Michael Sand, who is responsible for several central nervous system (CNS) clinical programmes at Boehringer Ingelheim. Participants are also asked to describe certain images with emotional context.

The members of Ribeiro's research group, use a specifically designed research software to evaluate the recordings of these conversations. “Studies of schizo phrenia patients have shown that the illness is reflected in their speech patterns,” Sand remarks. A change in intonation and a decrease in the level of complexity in their speech may be early signs of the disease.

Boehringer Ingelheim and the research group now plan to use these findings to support early diagnosis of the illness. “In people predisposed to this illness, the onset of schizophrenia typically occurs no earlier then during adolescence”, Ribeiro comments. He adds: “Schizophrenia is an illness which gets worse as time goes by”. But if it is diagnosed early and properly treated, the process may be slowed down or even stopped. 

Ribeiro is optimistic that doctors will in the near future already be able to make reliable predictions for risk groups – with the aid of the intelligent software he also uses in his research. While it used to take days or weeks to analyse speech patterns, now all that it requires is a small number of clicks and a few seconds. “We have become much faster thanks to digitalisation,” says Ribeiro. “Hopefully this will soon benefit a lot of people who are at risk of developing the illness.”